Negative Externalities of Irrigation Infrastructure: Forests, Farms, and Fertilizers in Palo Verde, Costa Rica

Rodrigo A. Arriagada, Erin O. Sills, Subhrendu K. Pattanayak, Frederick Cubbage and Eugenio González

Abstract The natural ecosystem of Palo Verde in Costa Rica is threatened by private land uses and public policies in regions upstream from the region. Irrigated rice farms rely on agrochemicals that pollute wetlands ecosystems in and near Palo Verde. Duality theory was used to estimate cost functions for farmers in the Palo Verde watershed to assess the influence of key production factors on rice production, and the subsequent off-site impacts on the natural ecosystem. The results indicate that rice seed and fertilizer prices had the most impact on fertilizer use, while land area farmed, irrigation cost, family size, and land preparation method did not significantly affect costs. Future policies should consider a mix of compensation payments to reduce fertilizer use and price incentives or taxes to enhance the coexistence of people and parks by encouraging farming practices that do not threaten ecosystems downstream.

1.

Introduction Rice is a basic ingredient of the Costa Rican diet. Its cultivation started as a

subsistence activity by small farmers that involved the use of local rice varieties highly susceptible to pests and with low response rate to fertilizers (Pérez, 2002). However, its dynamic changed with introduction of new seeds, technical assistance, and better cultivation techniques. Now rice cultivation is much more intensive, especially in the areas with availability of irrigation water. For example, the Arenal-Tempisque watershed in northwestern Costa Rica is one of the most economically productive regions of the country. The irrigation district associated with this watershed is also the largest in the country making this region the premier producer of rice. Since the establishment of the Arenal-Tempisque Irrigation Project or Proyecto Regional Arenal-Tempisque (PRAT) as it is known by its Spanish acronym, farmers have enjoyed the benefits of irrigation. 1

PRAT, establishment of the Arenal Forest Reserve, and trade protection for domestic rice production have all contributed towards a shift from traditional to intensive irrigation-based rice production, and the concomitant sharp increase in agrochemicals use. According to Espinoza-Esquivel and Arrieta-Espinoza (2007), rice production in Costa Rica faces several constraints, including weed control, among others. Full costaccounting of such off-site externalities is one step to analyze such impacts. However, policy analysis must go beyond accounting to identifying causes and consequences of private choices, if we are to identify the economic drivers for land use and policy levers for ecological conservation. Such information is lacking about rice cultivation, local economy and environment in this region of Costa Rica. Specifically, do not have empirical analysis of demand for agrochemicals such as pesticides, herbicides and fertilizers. Accordingly, the objective of this study was to perform a sound empirical analysis of market factors that affect rice production in the region in order to determine how market-based policy instruments could be used to reduce negative impacts of farm production on the adjacent natural wetlands. This study estimated a fertilizer demand function to explain the factors that influence the amount of fertilizers farmers use to cultivate rice in three agricultural communities around Palo Verde. Our model of fertilizer demand recognizes that input and output prices, fixed factors, and soil conditions may vary among farmers and communities. By providing information on how markets and other farm factors affect production of rice in the region, we assess how to incorporate the concept of protection of environmental services in the context of irrigation projects sustained by natural watersheds.

2.

Background For decades, the economy of the Tempisque region has long been based on

extensive cattle ranching (Peters, 2001). With the implementation of PRAT, commercial, crop-based production has become more important than cattle ranching in the lower basin, and flooded rice, sugar cane and melon cultivation are increasingly prominent features of the lower Arenal watershed (Daniels and Cumming, 2008). In particular, 2

PRAT has increased the reliability of the water supply, which in turn enabled farmers to move from one to two crops per year, resulting in a significant increase in land productivity and contributing to national food security (Hazell et al., 2001). The rivers in PRAT have their headwaters in the cloud forest of Arenal, and thus both this irrigation project and the country’s hydroelectricity rely on upstream watershed protection. Currently, PRAT covers approximately 28,000 ha and benefits approximately 1,125 families who produce mainly sugarcane, fodder, rice and fish from fish farms (400 ha of ponds) (Ballestero et al., 2007). The shift from traditional to intensive irrigated farming has led to a sharp increase in the use of fertilizers and pesticides. This situation has resulted in increased levels of nitrates, phosphates and other chemicals in the drainage water (Hazell et al., 2001). This situation is especially sensitive around the Palo Verde National Park, downstream from the Arenal-Tempisque irrigation district. Palo Verde is simultaneously a spectacular wetland, one of the most ecologically diverse parks in the country, and perhaps one of the best examples of tropical dry forests remaining anywhere in the world (Daniels and Cumming, 2008). The loss of soils and the water contamination appear to be the most critical negative impacts of the intensive use of agrochemicals on Palo Verde (Mc Coy, 1998). The problem is aggravated by the continued use of early high-yielding varieties of rice that have poor pest resistance and need more frequent spraying (Hazell et al., 2001). These soil sediments and agricultural runoff in the drainage water are contributing to the degradation of the wetlands located inside the park that is also highly recognized for being one of the biggest bird sanctuaries in Central America. Additionally, Daniels (2004) found that the major drivers of wetland conversion in the Tempisque Basin were related to agricultural intensification in the lower watershed. Declining water quality from an increase in chemical and soil sediment loading has become a major concern for the Costa Rican government in recent years. Water quality concerns have been considered as part of the Inter-American Development 3

Bank’s loan for further infrastructure construction as well (Hazell et al., 2001). These concerns have prompted more effort to quantify biological impacts of runoff on the watershed. Our study extends these analyses to examine economic factors that drive agrochemical use, and policy responses that could be used to protect the watershed in the face of increased rice production.

3.

Conceptual model Prior economics research has developed direct and indirect approaches for the

estimation of input demand and output supply (Chembezi, 1990). Indirect approaches include deriving demand functions from agronomic response functions and research. Direct methods include estimating demand functions directly from observed market data on fertilizer consumption and price, as well as other exogenous variables such as prices of farm products. A common economic alternative is to employ duality theory. Lau and Yotopoulos (1972) proposed a measurement of relative economic efficiency based on the concept of the profit function, as first introduced by McFadden (1970). This approach allows us to derive supply and factor demands without direct estimations of a production function, which is especially difficult when there is missing data, non-linearities and degrees of freedom problems (e.g., number of parameters to be estimated). Thus, factor demands can be derived from profit maximization as well as from cost minimization of the firm. Conditional factor demands are a collection of functions that give the optimal demands for each of several inputs as a function of the output expected, and the prices of inputs. In essence, the duality theory of production rests on the premise that cost function and production function contain the same information about the underlying production technology (Ahammad and Islam, 1999). The conditional factor demand approach was applied to explain fertilizer consumption in this research. We assumed that the production function is given by: , ,…,

;

4

,…,

,

(1)

where Y is the rice output, and X and Z are the vectors of variable and fixed inputs respectively. Conditional factor demands give the cost minimizing amount of every input as function of their prices and quantity of output produced. By Shepard’s Lemma, the firm’s system of cost minimizing input demand functions (the conditional factor demands) may be obtained by differentiating the cost function with respect to input prices: ,

,

,

(2)

where Xi is the quantity of input i used at the vector of factor prices, W, when y units of output are made using the least costly combination of factors. The parameters of the fertilizer demand function are estimated by single equation Ordinary Least Squares (OLS) procedure. Linear least squares regression provides information regarding the small-sample accuracy and precision of estimates, which may not be available from other techniques (Griffin et al., 1987). Statistical concerns regarding simultaneity of fertilizer demand and price are alleviated by the fact that fertilizer prices in Costa Rica are announced in advance of the planting season. Thus, farmers base their decisions about input quantities on the price announcements, which in turn are unaffected by local demand in the study area within the same period.

4.

Study area and data collection Figure 1 shows a map of Costa Rica and the study area where rice plots from the

communities included in the study are highlighted. The study is located in the Province of Guanacaste, a part of the Chorotega Region that is in the north of Costa Rica. The province has approximately 264,000 inhabitants, with 42% and 58% in urban and rural areas respectively. The agriculture sector has the largest number of workers with 28% of the economically active population which accounts for 90,395 people.

5

This region is particularly important due to its economic activities. Tourism, aquaculture, sugar cane and rice plantations have brought prosperity and employment to the area. Specifically, the Guanacaste province is the main rice production zone of the country. During the period 2006-2007, 18,304 ha were cultivated with rice in the Chorotega region which represents 39% of the total area planted in Costa Rica in those years (CONARROZ, 2008). The Palo Verde National Park is located on the Pacific slopes of Guanacaste. It includes 20,000 ha with a seasonally dry forest on limestone outcrops and extensive wetland vegetation bordering the Tempisque River that flows into the Gulf of Nicoya. The Tempisque and Bebedero watersheds represent 10% of the country, 53% of the Guanacaste Province. The Tempisque river basin has been the site of important biophysical, productive and social transformations over the last five centuries, which have shaped it into a complex matrix of agricultural lands, wetlands, protected areas and human settlements (Jiménez et al., 2001). Small farmers, beneficiaries of the state land reform that provided land of mainly rice-growing plots, own 7 to 10 ha parcels. Collectively, they represent more than 50% of the current PRAT irrigated area. The Institute for the Agrarian Development (IDA), a governmental institution, has promoted the formation of human settlements on rural locations. The farmers that live in the communities and were included in this survey have received their lands from IDA on the condition that they live in the settlement. The farmers receive their titles according with the Costa Rican legislation, which does not allow owners to sell the land. IDA selects the farmers that can receive the land according with a list of requirements that each applicant must meet, including (a) the obligation of being married, and (b) without previous land ownership. We selected five agricultural communities located around Palo Verde that collectively have a population of 336 households. The farming system used in three of these communities (Bagatzi, Falconiana and La Soga) is representative of the system used by other farmers in the Guanacaste Province. Farmers of the other two communities 6

(San Ramón and Playitas) use a farming system that is representative of the farming system used by private companies on that region. The data for our study were collected through structured interviews conducted by the lead author with the heads of farm households in June and July of 2003. Our survey instrument was adapted from the agriculture module of the Living Standards Measurement Study (LSMS), including LSMS surveys administered previously in countries similar to Costa Rica (e.g. Nicaragua and Panama). The questionnaire was also reviewed by two local technicians based at the Organization for Tropical Studies (OTS) in Palo Verde who have extensive field experience working with rice farmers in the area. The average farm size included in the study was 8.6 ha. San Ramón and Playitas had the biggest average farm size (11.7 ha) whereas Falconiana had the smallest farm size with 6.5 ha (Table 1). 90% of the land in these communities receives irrigation from PRAT. Of the total land area included in this study, 518 ha (71%) are covered by rice. This percentage varies across the five communities: Bagatzi 260.3 ha (79%), Falconiana 78.5 ha (68%), La Soga 112.5 ha (71%), and San Ramón and Playitas 66.8 ha (52%). The rest of the area is covered mainly by sugar cane, watermelon and pasture. San Ramón and Playitas were considered together because of their proximity and use of more mechanized rice production system compared with the other three communities of similar rice production techniques. Use of chemicals is intensive and their application is highly mechanized with the use of airplanes and tractors. The machinery (tractors, airplanes, and sewer machines) is provided by a local private company that works with these farmers through an agreement (land owners cede their lands to the company during the rice season and farmers receive a proportion of the harvest revenue after covering all the costs by the company). The large scale ground and aerial machinery requires that farmers cooperate to have adjacent land treated at the same time, and most farmers cooperate to contract these services. Labor is supplied by the private company and sometimes the owner of the land is not involved at all; but other times, especially for the maintenance activities, the farmer can be hired to work on his or her own land. This is a labor agreement that is usually established at the beginning of 7

every season. The household survey implemented in these two communities did not show any variation in the rice production system and so the data from these two communities were dropped during the estimation of the fertilizer demand function.

5.

Empirical Results

5.1.

Fertilizer Demand Function Production data from 40 farmers in the communities of Bagatzi, Falconiana and

La Soga are included in the analysis. The variables required to estimate the fertilizer demand function were input prices (seeds and fertilizer), output (rice) production, and quantities of fixed factors (land, irrigation cost, family size, and land preparation). Table 2 shows the descriptive statistics of the variables included in this analysis. The coefficients of variation suggest sufficient variation across farmers to permit statistically significant parameters to be estimated. The high variation in land area is explained by the presence of farmers that rent land from other farmers and therefore are able to cultivate larger areas (rental payments were not subtracted from profits and rental land is not included in Z1). Land preparation varies depending on the more or less intensive use of tractors during the land preparation (farmers use a technique called fangueo, or mechanical soil mixing to improve soil structure and eliminate weeds). Personal communications with several specialists in this area suggested that Falconiana is characterized by higher soil quality. Thus, a dummy variable for Falconiana was included in Table 2 to test for differences in fertilizer use. The irrigation cost varies because some farmers do not pay on time and pay a late fee. If farmers pay their irrigation bills on time they face a fixed cost, however if farmers do not keep track of their due dates to pay their bills they face higher costs. Then, Z2 can be interpreted as a proxy for financial management skills. The variation in cost of land preparation is mainly due to differences in the use of technology (e.g. tractor vs. ox), then Z4 can be treated as a index of capital intensity of technology. Given the size of our data, we rely on the traditional form of empirical production economics and assume a Cobb-Douglas functional form: 8

,

0,

1,2, … , ,

(4)

where Xi is a vector of inputs represented by rice seed and fertilizer price, land, irrigation payment, family size, land preparation, and the community dummy. By applying the laws of duality, the empirical specification of the fertilizer demand is given by,

ln

ln

ln

ζi ln

,

(3)

where F is the amount of fertilizer used in quintals per hectare; Dc is a dummy variable for farmers located in Falconiana, Y is the rice output (quintals/ha); W1 is the price of rice seeds; W2 is the price of fertilizer; Z1 is land; Z2 is irrigation payment; Z3 is family size; Z4 is cost of land preparation; α, β, γ, δ, and ζ are the estimated parameter coefficients; and ε is the error term. We should expect a positive impact of the community dummy (proxy for soil quality given the perception of higher soil fertility for land located in Falconiana), and a negative impact of fertilizer price. Following the cost minimization problem, output quantity (Y) should have a positive impact on fertilizer demand. Fixed inputs should have a positive impact (land and family size), and land preparation should negatively impact fertilizer use. The parameter estimates for the fertilizer demand function are shown in Table 3. The estimated fertilizer demand function is significant at the 1% level and explains almost 46% of the variation in the use of fertilizers among farmers that cultivate rice around Palo Verde. The estimated coefficient for the fertilizer price is negative, which means that farmers tend to apply less fertilizer when the price is higher. Note that this variable is significant at the 15% level even considering the small size (40 observations) of this cross-sectional data set. Almost all the farmers said during the interviews that they are more or less insensitive to the price of fertilizers because they see this input as an 9

essential factor to get a good harvest at the end of the season. However, there is some evidence that farmers do tend to reduce the use of fertilizer when its price increases. The coefficient on rice seed price is positive and significant at the 1% level. The price of seed indicates quality, as varieties that are more resistant to pests are more expensive. According with conversations with local personnel and field technicians at the Organization for Tropical Studies in Palo Verde, the positive influence of seed price on fertilizer demand may be explained by a lack of information on the real fertilizer requirements of the rice varieties being planted (some farmers apply more fertilizer when buying more expensive seeds just because they think they will avoid potential harvesting losses and obtain more productive harvests). The community dummy is significant as anticipated, at the 15% level. The limited information on soils in this region shows that soil properties in Falconiana permit a higher intensity and diversified use. Field technicians have tested the soils and provided this information to farmers. Among the communities included in this survey, the fact that Falconiana has the greatest variety of crops (rice, watermelons, melons, onions, tomatoes and other vegetables) offers further evidence of better soil quality. The estimated coefficient for output quantity is also significant at the 15% level. An issue of central interest to policy makers is the responsiveness of output supply and factor demands to price policy, and how they are affected by investment and land polices (Adesina and Djato, 1996). Fertilizer demand is fairly inelastic with respect to its own price (-0.456), which makes sense given the few substitutes that farmers have for chemical fertilizers. This is an important finding when considering fertilizer price as a policy mechanism to discourage the use of fertilizers in this particular region of Costa Rica. These results are consistent with predictions of theory and previous studies. The cross price elasticity of fertilizer demand with respect to the price of rice seed is fairly elastic (1.908) which is consistent with the previous explanation of the possible relationship between quality of seeds and fertilizer demanded (mainly a problem of 10

information). The introduction of improved seeds varieties that need less fertilizer can help to reduce the use of it and therefore reduce the environmental impacts that these chemicals cause to the environment (especially on the wetlands of Palo Verde), but the introduction of these improved seeds must be accompanied with good information about the benefits of purchasing them (e.g. less intensive use of fertilizers).

5.2

Risk Analysis and Policy Lessons The estimates of the fertilizer demand function presented in Table 3 show that

seed and fertilizer price, output quantity, and the community dummy were significant at explaining fertilizer demand. These results can be used to analyze future trends in fertilizer consumption under conditions of risk and uncertainty, and where probability distributions for input prices (seed and fertilizer) and output production can be defined in a way that best represent the data included in this study. Results of the fertilizer consumption simulation are presented in Table 4 where two community scenarios were defined to test the influence of the community dummy presented in Table 3. These results are based on the forecast range between range minimum and maximum arbitrarily defined on the field data collected during the interviews. The certainty levels represent the probability of using a quantity of fertilizer greater than the current average use in this region (11 quintals/ha).1 According to Table 4, both community scenarios forecast a 30% probability of using an amount of fertilizer greater than the current use in this region which can have important implications for the wetlands conservation in Palo Verde. Under this situation, different policy scenarios can be defined as summarized below in order to reduce fertilizer demand in a predicted situation where persistent high levels of fertilizer consumption are forecasted.

i.

Command and control: mandated reduction in fertilizer use

If we consider a policy scenario where the Costa Rican Government directly imposes a reduction in fertilizer consumption, important implications can be derived from 1

More details on risk analysis and simulation results can be found in Arriagada (2004). 11

the experiences in other countries and the current social and economical situation of Costa Rica. Previous experiences have concluded that due to high costs associated with enforcement and monitoring imposing a quantity restriction may not be feasible.

ii.

Fertilizer tax: raising the price of fertilizer

In this policy scenario, we are assuming the imposition of a tax on fertilizer consumption, with the purpose of discouraging its use. According with results of Table 3, a 1% change in fertilizer price leads to a -0.5% change in fertilizer consumption. Changing fertilizer price also have an impact on profitability of rice production. As we did with the estimation of fertilizer demand, we can also estimate a profit function depending on input prices and fixed inputs. Table 5 shows the results of a normalized profit function for rice production around Palo Verde. According with Table 5, a 1% change in fertilizer price leads to a -0.7% change in profits. If the Costa Rican Government seeks to discourage fertilizer use, Table 6 shows different levels of fertilizer price as a result of the tax and the corresponding reduction in fertilizer consumption and profits based on results presented in Table 3 and 5. According to Table 6, the imposition of a tax on fertilizer use has more impact on profits than on fertilizer consumption given the few substitutes for fertilizer farmers have in this area. Different targets of reduction in fertilizer use leads to a different levels of taxation. The upper bound on the loss of profits is estimated based on the data collected in the field (average farmer profits).

iii.

Rice production quota: limiting production of rice

Another alternative to indirectly discourage the use of fertilizer in rice production around Palo Verde is through a restriction on rice production (quotas on output production). According to Table 3, a 1% change in output quantity leads to a 0.4% change in fertilizer consumption. If we consider this relationship, we can estimate the economic impact of the implementation of various reductions in rice production (quotas). Table 7 presents different production quotas that lead to various levels of reduction in fertilizer consumption and the upper bound on the loss of revenues (given by the 12

economic valuation of the reduction in output quantity for an average rice farmer of this area). In this case, if the Costa Rican government wants to reduce the use of fertilizer by 15%, a quota on rice production of 40% has to be established. The same result in reduction of fertilizer use can be achieved by imposing a tax of 29% on fertilizer price. Results of Table 7 can be interpreted in two ways: (a) the economic valuation of the reduction in rice production can be considered as a subsidy that farmers should receive from the government for reducing fertilizer consumption, or (b) the economic valuation of the reduction in rice production can be considered as the amount farmers should pay for the environmental disservice fertilizer produces on the wetlands of Palo Verde (in this case farmers pay to avoid the reduction in fertilizer use). Under this policy scenario, rather than apply a tariff, the government could chose to directly subsidize rice production. If it limited the amount of production it subsidized, this would essentially act like a quota – or a quantity control – on rice production. Results of Table 7, suggest that depending on the level of reduction in fertilizer use we want to get, the payments go from USD 75 per hectare to USD 375 per hectare. In this case, instead of valuing the environmental services of the cloud forest in the upper watershed, we are valuing the environmental disservices on the wetlands of Palo Verde. Another consideration is the fact that this rice cultivation is made thanks to the Arenal-Tempisque Irrigation Project. Cloud forests on the upper Arenal-Tempisque watershed make this irrigation possible, which are another consideration that we are not taking into account in this analysis. In deciding, for example, a payment for environmental disservices cause for the use of agrochemicals, we are not including any payment for the environmental services provided by these cloud forests. Although we have been including irrigation payments in this study, these payments are basically for maintenance of the irrigation system and are not invested in direct environmental protection according to the information collected about it.

13

6.

Discussion The estimation of the fertilizer demand function suggests that price of fertilizer

and price of rice seeds are the variables that most affect the level of fertilizer demanded by farmers around Palo Verde in Costa Rica, as one expect. Results also suggest that farmers of Falconiana tend to demand less fertilizer, perhaps due to better soil quality in this community. The estimated function explains 46% of the variation in the quantity of fertilizer demanded. Other factors of production analyzed did significantly affect the cost functions estimated, including land area, irrigation cost, family size, or land preparation method. More data might find these factors to be more important, but they were not significant at all in this research. The price elasticity of fertilizer demand was –0.456 and the cross price elasticities with respect to price of rice seeds was 1.908. The use of fertilizers has an important impact on rice yields for landowners where a 1% change in the use of fertilizer leads to a 3% change in output quantity. With more variation in output price, unconditional factor demands could be estimated to determine the elasticity of fertilizer demand with respect to the price of rice. Different policy scenarios seeking the reduction in fertilizer use were analyzed. The direct intervention on fertilizer price through the imposition of a tax had the greatest impact on fertilizer demand. The forecast values of future fertilizer consumption provided evidence to conclude that the consumption of fertilizer will keep its high level. Policies to reduce fertilizer consumption should consider two important elements, the impact of the fertilizer on rice yields in this particular region of Costa Rica and the environmental impact of its use on the wetlands of the Palo Verde National Park. The results suggest that the introduction of new varieties of rice seeds could help to improve the level of profitability and decrease the intensive use of fertilizer. The development of new seeds varieties that are more pest resistant and more productive is being pursued in many other countries that produce rice. Better seed varieties with 14

decreasing marginal response to fertilizer can promote less intensive use of these agrochemicals. This results is consistent with previous studies (see Adamson, 2006; Bermúdez and Víquez, 2006; Espinoza-Esquivel and Arrieta-Espinoza, 2007) where the introduction of improved rice varieties would imply a greater flexibility in the management of the crop; a reduction in the frequency and amount of agrochemical applications; a reduction in the amount of agrochemical residues in the soil and reduction of production costs. It is worth noting that the Arenal-Tempisque Irrigation Project, the largest infrastructure project of its kind in Central America, is dependent on the protection of cloud forests in the upper watershed. Policy analysis can help assess farm promotion and production versus environmental protection, and go beyond accounting to identify causes and consequences of private choices. Irrigation and trade protection for domestic rice production have both contributed towards a shift from traditional to intensive irrigationbased rice production, with an increase in the use of fertilizers. Future policies should consider a mix of compensation payments to reduce fertilizer use and price incentives or taxes to enhance the coexistence of people and parks by encouraging farming practices that do not threaten ecosystems downstream. Our study provides further evidence of the tradeoffs between agriculture production and environmental protection, and helps identify the magnitudes of costs that would be required in this keystone area in Costa Rica. This conclusion is consistent with Daniels and Cummingan (2008) that studied wetland conservation in Northwest Costa Rica. They concluded that wetland conservation will have to encompass a regional perspective on ecosystems management within the mosaic of different land uses that influence environmental protection.

15

7.

Bibliography

Adamson, M. Diagnosis and proposal of the socio-economic impact of an introduction of genetically modified rice in Costa Rica. Paper presented at the Foro Internacional cultivos genéticamente mejorados y bioseguridad (Universidad de Costa Rica, San Pedro, 2006). Adesina, A. and Djato, K. Relative efficiency of women as farm managers: profit function analysis in Cote d’Ivore, Agricultural Economics, Vol. 16, (1997) pp. 47-53. Ahammad, H. and Islam, N. Estimating the agricultural production system of Western Australia: a profit function approach, Working Paper (Economic Centre, Department of Economics, University of Western Australia, 1999). Arriagada, R. Estimating profitability and fertilizer demand for rice production around the Palo Verde National Park, Costa Rica, Thesis (North Carolina State University, 2004). Ballestero, M., Reyes, V. and Astorga, Y. ‘Groundwater in Central America: its importance, development and use, with particular reference to its role in irrigated agriculture’, in M. Giordano and K.G. Villholth (eds). The Agricultural Groundwater Revolution: Opportunities and Threats to Development (CAB International, 2007, pp. 100-130). Bermúdez, S. and Víquez, M. Diagnosis and proposal for the environmental impact assessment of genetically modified crops (risk and benefits) in Costa Rica. Paper presented at the Foro Internacional cultivos genéticamente mejorados y bioseguridad. (Universidad de Costa Rica, San Pedro, 2006) Chembezi, D. Estimating fertilizer demand and output supply for Malawi’s smallholder agriculture, Agricultural Systems, Vol. 33, (1990) pp. 293-314.

16

Corporación Arrocera Nacional de Costa Rica (CONARROZ). Estadísticas año arrocero 2006-2007 (available at: http://www.conarroz.com/informacionestadistica.shtml, 2008). Daniels, A. Protected area management in a watershed context: a case study of Palo Verde National Park, Costa Rica, Thesis (University of Florida, 2004). Daniels, A. and Cumming, G. Conversion or conservation? Understanding wetland changing in Northwest Costa Rica, Ecological Applications, Vol. 18, (2008) pp. 4963. Espinoza-Esquivel, A. and Arrieta-Espinoza, G. A multidisciplinary approach directed towards the commercial release of transgenic herbicide-tolerant rice in Costa Rica, Transgenic Research, Vol. 16, (2007) pp. 541-555. Griffin, R., Montgomery, J. and Rister, M. Selecting functional forms in production function analysis, Western Journal of Agricultural Economics, Vol. 12, (1987) pp. 216-227. Hazell, P., Chakravorty, U., Dixon, J. and Celis, R. Monitoring systems for managing natural resources: economic indicators and environmental externalities in a Costa Rican watershed, EPTD Discussion Paper No. 73 (International Food Policy Research Institute and The World Bank, 2001). Jiménez, J., González, E., Mateo, J. Perspective for the integrated management of the Tempisque River Basin, Costa Rica, Working Paper (Organization for Tropical Studies, 2001). Lau, L. and Yotopoulos, P. Profit, supply and factor demand functions, American Journal of Agricultural Economics, Vol. 54, (1972) pp. 11-18. McCoy, M. La siembra de arroz con riego en lámina de agua: tres anos de experiencias empíricas en Guanacaste para evitar la erosión de suelos y el uso de herbicidas, Working Paper (Universidad Nacional de Costa Rica, 1998). 17

McFadden, D.L. Cost, Revenue, and Profit Functions, Working Paper (University of California, Department of Economics, 1970). Pérez, J. El cultivo del arroz en Costa Rica y en el proyecto de riego Arenal-Tempisque, Working Paper (Organización para Estudios Tropicales-Fundación Avina. Palo Verde, 2000). Peters, G. ‘La cuenca del Tempisque: una perspectiva histórica’, in J. Jiménez and E. González (eds). La cuenca del Rio Tempisque: perspectivas para un manejo integrado (Organización para Estudios Tropicales, 2001).

18

SECTION FIGURES

Figure 1. Map of the study area and location of rice plots

19

SECTION TABLES Table 1. Number of interviewed farmers and average size plot

*

Community Bagatzi Falconiana La Soga San Ramon and Playitas Totals

Number of interviews 22 12 13 11 58

Average size plot (ha)* 9.18 6.46 7.03 11.72 8.62

Total land area (ha)** 328.75 116.1 157.48 128.95 731.28

Only includes the average size of plots owned by each farmer Includes parcels of land owned and rented by each farmer

**

Table 2. Descriptive statistics of variables included in the fertilizer analysis Variable Y F W1 W2 Z1 Z2 Z3 Z4 Dc

Mean†

Description Rice output (quintals/ha×semester) Fertilizer demand (quintals/ha×semester) Price of rice seeds (Colones/quintal) Price of Fertilizers (Colones/quintal) Land (hectares) Cost of irrigation (Colones/ha×semester) Family size (number of people) Cost of land preparation (Colones/ha) Community dummy variable = 1 if farmer cultivates in Falconiana and 0 otherwise

106.9 10.3 9,769 4,864 10.8 8,959 4.7 27,939 0.20

Standard deviation 32.2 2.9 861 1,253 10.1 1,017 2.1 7,555 0.40

Coefficient of variation (%) 30.12 28.47 8.81 25.78 93.82 11.35 46.34 27.04 ---

1 quintal = 46 kilograms. 1 USD = 570 Colones † Including 40 households of Bagatzi, Falconiana and La Soga

Table 3. Fertilizer demand function for rice production around Palo Verde, Costa Rica Independent variables Intercept Y Output quantity W1 Seed price W2 Fertilizer price Z1 Land Z2 Irrigation cost Z3 Family size Z4 Land preparation Dc Community dummy N R2 Adjusted R2 F-value

***

Parameter α γ δ1 δ2 ζ1 ζ2 ζ3 ζ4 β 40 0.455 0.314 3.24

Estimated coefficient (OLS) -14.175 0.356* 1.908*** -0.456* 0.049 0.222 0.047 -0.095 -0.258*

Standard error

P-value

6.731 0.176 0.539 0.296 0.085 0.436 0.099 0.147 0.134

0.043 0.052 0.001 0.133 0.562 0.613 0.635 0.521 0.063

0.008

= 99% confidence, ** = 95%, * = 85%.

20

Table 4. Forecast parameter and statistics generated from Monte Carlo simulation (1,000 iterations) of fertilizer consumption around Palo Verde, Costa Rica Forecast

Unit

Parameter

Fertilizer consumption (5-year period)

Quintals†/ha

Mean Median St Dev CV (%) Range minimum Range maximum Certainty level

Community dummy scenario Bagatzi and La Soga Falconiana 10 11 5 6 10 10 100 90.91 1 1 36 36 32.99% 35.86%

†1 Quintal = 46 Kilograms

Table 5. Estimation of the normalized profit function for rice production around Palo Verde, Costa Rica Independent variables Profit function constant W1 Seed price W2 Fertilizer price Z1 Capital intensity Z2 Irrigation cost Z3 Family size Dc Community dummy N R2 Adjusted R2 F-value

***

Parameter ln A α1 α2 β1 β2 β3 δc 40 0.308 0.182 2.45

Estimated coefficient (OLS) 1.762 -2.375** -0.737 0.744* -0.755 -0.002 0.451

Standard error

P-value

9.839 0.951 0.661 0.457 1.083 0.227 0.344

0.858 0.017 0.272 0.112 0.490 0.989 0.199

0.045

= 99% confidence, ** = 95%, * = 85%.

Table 6. Economic impact of different levels of taxation on fertilizer consumption associated to rice production around Palo Verde, Costa Rica % change in fertilizer price (tax) 6% 10% 19% 29%

Implied % of reduction in fertilizer use 3% 5% 10% 15%

†1 USD 2009 (January, 20th) = 564 Colones

Implied % of reduction in profits 4% 7% 14% 21%

21

Upper bound on the loss of profits (Colones†/ha) 8,160 13,600 27,199 40,799

Table 7. Economic impact of different policies attempting to reduce the use of fertilizers in rice production around Palo Verde, Costa Rica % of reduction in output production (quotas) 8% 13% 26% 40%

Implied % of reduction in fertilizer consumption 3% 5% 10% 15%

†1 USD 2009 (January, 20th) = 564 Colones

22

Upper bound on the loss of revenues (Colones†/ha) 32,038 53,411 106,832 160,312