EFFECT OF SELECTED INSECTICIDE ON WHITEFLY (Bemisia tabaci) INFESTING BRINJAL PLANTS © 2015Asian Economic and Social Society. All rights reserved ISSN(P): 2304-1455/ISSN(E):2224-4433 Volume 6(2), 21-35 21 RICE PRODUCTION STRUCTURES IN SRI LANKA: THE NORMALIZED TRANSLOG PROFIT FUNCTION APPROACH Chatura Sewwandi Wijetunga PhD Candidate; Department of Agricultural and Resource Economics, The University of Tokyo, Japan Abstract 1 This study attempts to estimate the output supply and input demand elasticities of rice production using the restricted normalized translog profit function for the four major paddy producing districts; Anuradhapura, Hambantota, Kurunegala and Polonnaruwa in Sri Lanka. In addition, elasticities of substitution between inputs are also estimated. The results suggest that the changes in market prices of inputs and output significantly affect the farmers’ profits, rice supply and the use of resources in paddy cultivation. The supply elasticity of rice with respect to its own price is 0.5 and the supply elasticity of output with respect to fertilizer price is -0.05 on an average. Fertilizer demand in the country is inelastic but significant to its own price. Therefore, fertilizer subsidy is one of the main factors to increase fertilizer demand as well as paddy supply in the country. In addition, the low elasticity of substitution between labour and fertilizer and other inputs indicates that there is a complementary relationship among these inputs hence their combined application increases paddy production synergistically. Overall, this study suggest that farmers are price sensitive and assures prevalence of higher output price is essential for higher rice production. Keyword: Translog profit function, demand elasticity, supply elasticity, elasticity of substitution 1. INTRODUCTION 1.1. Background of the study The development of the agriculture sector is imperative to the wellbeing of the people since it still plays a key role in Sri Lanka in terms of employments, and main consumption expenditure. Rice is the main agricultural produce, like many other developing countries in Asia and country’s staple food, and it has long been an important national target for attaining rice self-sufficiency. In the year 2014, it contributed 10 percent to the agricultural GDP and 1.1 percent to the total GDP (Central Bank of Sri Lanka 2014). The production and average yield of paddy in the country has tremendous improvement after independence in 1948, is mainly attributed the incentives provided by successive government through various measures such as huge investments in irrigation infrastructure, settlement schemes, introducing high yielding varieties (HYV), fertilizer subsidy and other input and output subsidies to farmers. As a result, Sri Lanka is almost achieved self- sufficiency in rice in good crop years. In spite of significant advances in rice production sector and various incentives provided to the sector, there is a rising concern about escalating the cost of production and declining profitability in rice cultivation. Thiruchelvum (2005) also pointed out that Sri Lanka maintains high level of self-sufficiency in rice at a cost to the economy. In addition, the current fertilizer subsidy program seems highly cost ineffective due to government spends between 1.4 and 2.4 Rupees per acre to increase farm income by only Rupees 1 per acre (World Bank, Corresponding author’s Email address: chaturawijethunga@yahoo.com Asian Journal of Agriculture and Rural Development journal homepage: http://aessweb.com/journal-detail.php?id=5005 mailto:chaturawijethunga@yahoo.com mailto:chaturawijethunga@yahoo.com mailto:chaturawijethunga@yahoo.com mailto:chaturawijethunga@yahoo.com Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 22 2013). The current subsidy accounts 4 percent of agriculture GDP and 2 percent of the total government expenditure in 2014 (Central Bank of Sri Lanka, 2014). Therefore, Fertilizer subsidy is continuing as the most controversial input subsidy programme in Sri Lanka and the subsidy cost making a huge burden to the government budget. The main objective of this paper, therefore, is to analyze input demand and output supply parameters of paddy production with special reference to the fertilizer subsidy which has given significant interest in the recent years. In addition, study intends to estimate elasticities of the paddy supply and input demand which are vast important for the accurate prediction of the responsiveness of farmers to changes in input-output prices. Further, our analysis also aims to estimate the elasticities of substitution among different inputs used in paddy cultivation in Sri Lanka. In order to meet the above objectives and to address the remainder of this paper is arranged as follows. Next section presents brief review of literature focusing various econometric techniques employed for the estimation of output supply and input demand functions. Section 2 presents the methodology employs to achieve the intended objectives. Section 3 discusses the results using normalized translog profit function approach that jointly estimates output supply and input demand parameters in rice production. The final section presents the concluding remarks. 1.2. Literature review There are various econometric techniques have been employed for the estimation of demand for inputs and supply of agricultural crops. Among them production function approaches, cost function approach, profit function approach etc. are widely used. Direct or indirect application of Cobb- Douglas production function is based on highly restricted assumptions of unitary elasticity of substitution, constant returns to scale hence it yields invalid elasticities (Diewert, 1971; Christensen et al., 1973). In addition, production functions with Constant Elasticity of Substitution (CES), Variable Elasticity of Substitution (VES) and the nested CES production functions are applied to estimate the production structures. However, they are based on rigid restrictions and incapable of explaining exact relationships among variables (Chaudhary et al., 1998). Consequently, the duality approach was widely applied to provide a comprehensive relationship among inputs and outputs (Beccera & Shuway, 1992; Siregar, 2007). The translog form is flexible because specific features of technology such as returns to scale or homotheticity may be tested by examining the estimated model parameters (Ray, 1982). Berndt and Christensen (1973), Berndt & Wood (1975), Christensen et al. (1975) and Binswanger (1974) also employed translog models. In addition, many authors have used profit function in empirical estimation of factor demand and output supply parameters (Yotopoulos et al., 1976; Sidhu & Baanante, 1979 and 1981; Bapna et al., 1984; Ball, 1988; Fulginiti & Perrin, 1990; Altemeier & Bottema, 1991). Among the various functional forms a flexible functional form for the profit function is preferred (Diewert, 1973; Fuss et al., 1978; Lopez, 1985) and translog, normalized quadratic and generalized Leontief are some of them. Restricted normalized translog profit function is utilized as it is able to depict input demand and output supply simultaneously. The translog profit function is a flexible functional form to estimate the input demand as it can eliminate problems related to the restrictive as required by Cobb-Douglas profit function (Diewert, 1971). Therefore, many researchers chose to start from a profit function and derive input demand and supply response functions from the profit function based on Hotelling’s Lemma (Wall & Fisher, 1988; Hattink et al., 1998). So far, in Sri Lanka only one study (Rajapaksha & Karunagoda, 2009) has examined the relationship among the multiple inputs used in paddy cultivation by applying the translog profit function approach using time series data. Despite that there were some previous studies which estimated the demand for fertilizer in Sri Lanka using various approaches. As example, Thusiman et al. (1987) estimated short run production elasticities by using micro level data and concluded that it is relatively less sensitive to increase paddy production in response to decrease in the price of Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 23 fertilizer. Meanwhile, a partial equilibrium model developed by Weerahewa (2004) to capture the changes in policy framework in demand and supply function of paddy has estimated the demand and supply elasticities for the period of 1978-2000. In that model, rice demand is a function of rice price, other crop prices and expenditure while supply is a function of paddy price and only two inputs: seed price and fertilizer price. In addition, Ekanayake (2006) has estimated the fertilizer demand elasticities for the period of 1981-2004 based on econometric method. According to his estimates fertilizer demand was relatively inelastic with respect to fertilizer prices. While, Weligamage et al. (2009) has estimated rice production functions in Kirindioya area using household survey data. Recently, Weerasooriya and Gunaratne (2009) examined the link between the changes in productivity and fertilizer use associated with the fertilizer subsidy using supply and area response functions for the period of 1983-2007. However, none of these studies have examined the elasticity of substitution of inputs used in paddy in Sri Lanka. Also there are no proper elasticity estimates for input demand and output supply in rice using recent input output data. Hence, our study seems to fill the knowledge gap in Sri Lanka using a normalized translog profit function approach. 2. METHODOLOGY This study estimates the input demand and output supply functions in four selected districts in Sri Lanka for the period of 1990-2012. The selected districts are Anuradhapura, Hambantota Kurunegala and Polonnaruwa which contributed around 11%, 6%, 11% and 13% correspondingly (average of 2009-12 periods) (Department of Census and Statistics, 2012) to the national rice production annually. Anuradhapura, Hambantota and Polonnaruwa districts belong to the dry zone where paddy cultivation is mainly under the irrigation schemes while Kurunegala district belongs to the Intermediate zone where paddy cultivation is under the rain–fed and irrigated schemes. In addition, commercial level paddy cultivation is conducted in all four districts. The demand for a production input is a derived demand based on the demand for final products. Farmers are assumed to behave rationally and the general profit function can be expressed as: 𝜋 = 𝑃𝑄 − 𝑤𝑥 Production function is given by: 𝑄 = 𝑓 𝑥, 𝑧 Where 𝜋 is profit, 𝑃 and 𝑤 are prices of output and inputs respectively. Output quantity is𝑄, while 𝑥 and z are the vectors of variable input quantities and fixed factor quantities. Therefore, profit function can be solved for the maximization situation. Max 𝑃𝑄 − 𝑤𝑥 subject to 𝑄 = 𝑓 𝑥, 𝑧 The solution for this problem is a set of input demand and output supply functions given by, 𝑥 = 𝑥(𝑃,𝑤, 𝑧) 𝑄 = 𝑞(𝑃,𝑤, 𝑧) Substituting the above equations in general profit function gives profit maximization level of input and output. 𝜋 = 𝑃′𝑞 𝑃,𝑤, 𝑧 − 𝑤 ′𝑥(𝑃,𝑤, 𝑧) Inverse input demand function and output supply functions can be obtained by differentiating the profit function with respect to the input price 𝑤 and output price𝑃. 𝑋𝑖 ∗ = − 𝜕𝜋 𝜕𝑤𝑖 = 𝑋∗(𝑃,𝑤, 𝑧) and 𝑄∗ = 𝜕𝜋 𝜕𝑃 = 𝑄∗(𝑃,𝑤, 𝑧) Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 24 A generalization of the Normalized trans-log profit function for a single output is given by Diewert (1974), Christensen et al. (1973). ln𝜋∗ = 𝛼0 + 𝛼𝑖 𝑛 𝑖=1 𝑙𝑛𝑃𝑖 ∗ + 1 2 𝛾𝑖ℎ 𝑙𝑛𝑃𝑖 ∗𝑛 ℎ=1 𝑛 𝑖=1 𝑙𝑛𝑃ℎ ∗ + 𝛿𝑖𝑘 𝑚 𝑘=1 𝑛 𝑖=1 𝑙𝑛𝑃𝑖 ∗𝑙𝑛𝑍𝑘 + 𝛽𝑘 𝑚 𝑘=1 𝑙𝑛𝑍𝑘 + 1 2 ∅𝑘𝑗 𝑚 𝑗=1 𝑚 𝑘=1 𝑙𝑛𝑍𝑘 𝑙𝑛𝑍𝑗 + 휀𝑖 ............ (1.1) Where, 𝛾𝑖ℎ = 𝛾ℎ𝑖 , 𝛿𝑖𝑘 = 𝛿𝑘𝑖 and ∅𝑘𝑗 = ∅𝑗𝑘 for h, i and k and the function is homogeneous of degree one in prices of all variable inputs and outputs. The definition of the variables and notation used in the profit function are as follows: 𝜋∗ is the restricted profit (total revenue less total cost of variable inputs) normalized by 𝑃𝑞 , the price of output, 𝑃𝑖 ∗ is the price of variable input 𝑋𝑖 , normalized by 𝑃𝑞 , 𝑍𝑘 is the k th fixed inputs; 𝑖 = ℎ = 1,… .𝑛 , 𝑘 = 𝑗 = 1,…𝑚; 𝑙𝑛 is the natural logarithm; and 𝛼0,𝛼𝑖 , 𝛾𝑖ℎ , 𝛿𝑖𝑘 , 𝛽𝑘 and ∅𝑘𝑗 are the parameters to be estimated and 휀𝑖 is random error. The partial derivatives of restricted profit function with respect to logs of input price yield the share equations as follows; 𝑆𝑖 = 𝑃𝑖 ∗𝑋𝑖 𝜋∗ = 𝜕𝑙𝑛 𝜋∗ 𝜕𝑙𝑛 𝑃𝑖 ∗ = 𝛼𝑖 + 𝛾𝑖ℎ 𝑙𝑛𝑃ℎ ∗𝑛 ℎ=1 + 𝛿𝑖𝑘 𝑚 𝑘=1 𝑙𝑛𝑍𝑘 ................... (1.2) 𝑆𝑞 = 𝑃𝑞 ∗𝑋𝑞 𝜋∗ = 1 + 𝜕𝑙𝑛 𝜋∗ 𝜕𝑙𝑛 𝑃𝑞 ∗ = 1 − (𝛼𝑖 + 𝛾𝑖ℎ 𝑙𝑛𝑃ℎ ∗𝑛 ℎ=1 + 𝛿𝑖𝑘 𝑚 𝑘=1 𝑙𝑛𝑍𝑘) .................. (1.3) Where 𝑆𝑖 is the share of 𝑖th input and𝑆𝑞 is the share of output (𝑞). 𝑆𝑞 is equivalent to the ratio of the total value of output to restricted profit. Since the output and input shares come from singular system of equations, their summation is equal to one and one of the share equations can be ignored. The normalized input prices and quantities of fixed factors are considered the exogenous variables under the price taking behavior. Using Hotelling Lemma, the translog profit function can be served to obtain the following share equations: Derived factor demand function: 𝑋𝑖 = − 𝜋∗ 𝑃𝑖 𝛼𝑖 + 𝛾𝑖ℎ 𝑙𝑛𝑃ℎ ∗𝑛 ℎ=1 + 𝛿𝑖𝑘 𝑚 𝑘=1 𝑙𝑛𝑍𝑘 .................. (1.4) Derived output paddy supply function: 𝑋𝑞 = 𝜋∗ 𝑃𝑞 [1 − (𝛼𝑖 + 𝛾𝑖ℎ 𝑙𝑛𝑃ℎ ∗𝑛 ℎ=1 + 𝛿𝑖𝑘 𝑚 𝑘=1 𝑙𝑛𝑍𝑘)] .................. (1.5) By using share equations and estimates of the profit function, output supply elasticity and input demand elasticities will be estimated simultaneously. Estimation of elasticities The elasticities of variable input demands and output supply with respect to all exogenous variables evaluated at averages of the 𝑆𝑖 and at given levels of variable input prices. These are the linear transformations of parameter estimates of the profit function. From (1.2) the demand equation for the 𝑖th variable input can be written as, 𝑋𝑖 = 𝜋 𝑃𝑖 ln − 𝜕𝑙𝑛𝜋 𝜕 ln 𝑃𝑖 .................. (1.6) 𝑙𝑛𝑋𝑖 = 𝑙𝑛𝜋 − 𝑙𝑛𝑃𝑖 + ln − 𝜕𝑙𝑛𝜋 𝜕 ln 𝑃𝑖 ................. (1.7) Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 25 The own- price elasticity of demand (𝜂𝑖𝑖 ) for 𝑋𝑖 is, 𝜂𝑖𝑖 = 𝜕 ln 𝑋𝑖 𝜕 ln 𝑃𝑖 = 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 − 1 + 𝜕𝑙𝑛 𝜕 ln 𝑃𝑖 − 𝜕𝑙𝑛𝜋 𝜕𝑙𝑛 𝑃𝑖 ................. (1.8) 𝜂𝑖𝑖 = −𝑆𝑖 ∗ − 1 − 𝛾𝑖𝑖 𝑆𝑖 ∗ .................. (1.9) Where 𝑆𝑖 ∗ is the simple average of 𝑆𝑖 . Similarly from (1.7), the cross price elasticity of demand (𝜂𝑖ℎ) for input 𝑖 with respect to the price of ℎth input can be obtained: 𝜂𝑖ℎ = 𝜕 ln 𝑋𝑖 𝜕 ln 𝑃ℎ = 𝜕 ln 𝜋 𝜕𝑙𝑛 𝑃ℎ + 𝜕𝑙𝑛 𝜕𝑙𝑛 𝑃ℎ − 𝜕 ln 𝜋 𝜕𝑙𝑛 𝑃𝑖 ................. (1.10) 𝜂𝑖ℎ = −𝑆ℎ ∗ − 𝛾𝑖ℎ 𝑆𝑖 ∗ .................. (1.11) Where𝑖 ≠ ℎ. The elasticity of demand for input 𝑖(𝜂𝑖𝑞) with respect to output price, 𝑃𝑞 , can also be obtained from (1.7); 𝜂𝑖𝑞 = 𝜕 ln 𝑋𝑖 𝜕 ln 𝑃𝑞 = 𝜕 ln 𝜋 𝜕 ln 𝑃𝑞 − 𝜕 ln 𝑃𝑖 𝜕 ln 𝑃𝑞 + 𝜕𝑙𝑛 𝜕 ln 𝑃𝑞 − 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 .................. (1.12) 𝜂𝑖𝑞 = 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 𝑛 𝑖=1 . 𝜕𝑙𝑛 𝜕 ln 𝑃𝑞 − −1 − 𝛾𝑖ℎ 𝑆𝑖 ∗ 𝑛 ℎ=1 (−1) .................. (1.13) Where 𝑖 = 1,……… ,𝑛, ℎ = 1,… . .𝑛, 𝜂𝑖𝑞 = 𝑆𝑖 ∗𝑛 1=1 + 1 + 𝛾𝑖ℎ 𝑆𝑖 ∗ 𝑛 ℎ=1 .................. (1.14) The elasticity of demand 𝜂𝑖𝑘 for input 𝑖 with respect to the 𝑘th fixed factor 𝑍𝑘 is also obtained from (1.7): 𝜂𝑖𝑘 = 𝜕 ln 𝑋1 𝜕𝑙𝑛 𝑍𝑘 = 𝜕 ln 𝜋 𝜕 ln 𝑍𝑘 − 𝜕 ln 𝑃𝑖 𝜕 ln 𝑍𝑘 + 𝜕𝑙𝑛 𝜕𝑙𝑛 𝑍𝑘 − − 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 , .................. (1.15) 𝜂𝑖𝑘 = 𝛿𝑖𝑘 ln𝑃𝑖 𝑛 𝑖=1 + 𝛽𝑘 − 𝛿𝑖𝑘 𝑆𝑖 ∗ .................. (1.16) Output supply elasticities We evaluate the output supply elasticities with respect to output price, price of variable inputs and quantities of fixed inputs at averages of 𝑆𝑖 and at given levels of exogenous variables. It can also be expressed as linear functions of restricted profit function parameters. Equation for output supply 𝑞 can be written as (1.17) using the duality theory. 𝑞 = 𝜋 + 𝑃𝑖𝑋𝑖 𝑛 𝑖=1 .................. (1.17) By using (1.4), gives the equation; 𝑞 = 𝜋 + 𝜋 − 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 𝑛 𝑖=1 , or 𝑞 = 𝜋 1 − 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 𝑛 𝑖=1 ............... (1.18) Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 26 ln 𝑞 = ln𝜋 + 𝑙𝑛 1 − 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 𝑛 𝑖=1 .................. (1.19) Elasticity of supply with respect to the price of 𝑖th variable input is given by equation (1.20) where𝑖 = ℎ = 1,… .𝑛. , 𝜖𝑞𝑖 = 𝜕 ln 𝑞 𝜕 ln 𝑃𝑖 = 𝜕 ln 𝜋 𝜕 ln 𝑃𝑖 + 𝜕𝑙𝑛 𝜕 ln 𝑃𝑖 1 − 𝜕 ln 𝜋 𝜕 ln 𝑃ℎ 𝑛 ℎ=1 .................. (1.20) For the translog profit function: 𝜖𝑞𝑖 = −𝑆𝑖 − 𝛾ℎ𝑖 𝑛 ℎ=1 1 + 𝑆ℎ ∗𝑛 ℎ=1 .................. (1.21) The own price elasticity of supply 𝜖𝑞𝑞 and elasticity of output supply 𝜖𝑞𝑘 with respect to the fixed inputs 𝑍𝑘 is calculated by using following equations. 𝜖𝑞𝑞 = 𝑆𝑖 ∗𝑛 𝑖=1 + 𝛾𝑖ℎ 𝑛 ℎ=1 𝑛 𝑖=1 / 1 + 𝑆ℎ ∗𝑛 ℎ=1 ................... (1.22) 𝜖𝑞𝑘 = 𝛿𝑖𝑘 𝑛 𝑖=1 ln𝑃𝑖 + 𝛽𝐾 − 𝛿𝑖𝑘 𝑛 𝑖=1 / 1 + 𝑆ℎ ∗𝑛 ℎ=1 ................... (1.23) Partial elasticities of substitution The partial elasticities of substitution are normalized price elasticities, represents how factor income shares change as the ratio of the factors change. Atkinson & Halvorsen (1976) defined the Partial elasticities of substitution as: 𝜎𝑖𝑖 = 1 𝑆𝑖 𝜂𝑖𝑖 ................... (1.24) and 𝜎𝑖ℎ = 1 𝑆ℎ 𝜂𝑖ℎ ................... (1.25) Where 𝜎𝑖𝑖 is the own elasticity of substitution and 𝜎𝑖ℎ is the cross elasticity of substitution. 2.1. Data To estimate the model, labour, seeds, fertilizer and agro chemicals are included as variable factors of production of paddy. In addition the machinery cost and land are included as fixed factors of production. The profit function analysis is based on the data obtained from the biannual Cost of Cultivation Surveys conducted by the Socio Economics and Planning Centre of the Department of Agriculture for the period of 1990 to 2012. Weighted average price of urea, TSP and MOP are used as fertilizer price for the analysis. GDP deflator index values obtained from the Annual Reports of Central Bank are used as proxy to calculate the machinery price index and Real price index values are obtained compared to Year 2000 values. Total cost of each inputs are calculated for the districts by multiplying per acre input cost from the paddy extent harvested area in each districts obtained from the Department of Census and Statistics for the each year. Yala and Maha season cost data are aggregated to obtain the total annual cost. Farm gate price of paddy is obtained from the Department of Agriculture and one year lagged farm gate price is used for the model. All the time series variables are tested for the presence of stationary prior to perform the econometric analysis. In order to test for unit root at its level or first difference, all the variables are subjected to Augmented Dickey Fuller (ADF) unit root test. Since the variables of the time series are stationary their variances and auto covariances are independent of time. Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 27 2.2. Model estimation From the general function (1.1), the normalized restricted translog profit function can be specified in actual variables as: ln𝜋∗ = 𝛼0 + 𝛼𝐿 ln𝑃𝐿 ∗ + 𝛼𝑆 ln𝑃𝑠 ∗ + 𝛼𝐹 ln𝑃𝐹 ∗ + 𝛼𝐶 ln𝑃𝐶 ∗ + 1 2 𝛾𝐿𝐿 ln𝑃𝐿 ∗ ln𝑃𝐿 ∗ + 1 2 𝛾𝑆𝑆 ln𝑃𝑆 ∗ ln𝑃𝑆 ∗ + 1 2 γ FF ln𝑃𝐹 ∗ ln𝑃𝐹 ∗ + 1 2 γ CC ln𝑃𝐶 ∗ ln𝑃𝐶 ∗ + 1 2 γ LS ln𝑃𝐿 ∗ ln𝑃𝑆 ∗ + 1 2 γ LF ln𝑃𝐿 ∗ ln𝑃𝐹 ∗ + 1 2 γ LC ln𝑃𝐿 ∗ ln𝑃𝐶 ∗ + 1 2 𝛾𝑆𝐹 ln𝑃𝑆 ∗ ln𝑃𝐹 ∗ + 1 2 𝛾𝑆𝐶𝑙𝑛𝑃𝑆 ∗ ln𝑃𝐶 ∗ + 1 2 𝛾𝐹𝐶 ln𝑃𝐹 ∗ ln𝑃𝐶 ∗ + 𝛿𝐿𝑍𝑚 ln𝑃𝐿 ∗ ln𝑍𝑚 + 𝛿𝐿𝑍𝑎 ln𝑃𝐿 ∗ ln𝑍𝑎 + 𝛿𝑆𝑍𝑚 ln𝑃𝑆 ∗ ln𝑍𝑚 + 𝛿𝑆𝑍𝑎 ln𝑃𝑆 ∗ ln𝑍𝑎 + 𝛿𝐹𝑍𝑚 ln𝑃𝐹 ∗ ln𝑍𝑚 + 𝛿𝐹𝑍𝑎 ln𝑃𝐹 ∗ ln𝑍𝑎 + 𝛿𝐶𝑍𝑚 ln𝑃𝐶 ∗ ln𝑍𝑚 + 𝛿𝐶𝑍𝑎 ln𝑃𝐶 ∗ ln𝑍𝑎 + 𝛽𝑍𝑚 ln𝑍𝑚 + 𝛽𝑍𝑎 ln𝑍𝑎 + 1 2 ∅𝑍𝑚𝑍𝑚 ln𝑍𝑚 ln𝑍𝑎 + 1 2 ∅𝑍𝑎𝑍𝑎 ln𝑍𝑎 ln𝑍𝑎 + ∅𝑍𝑚𝑍𝑎 ln𝑍𝑚 ln𝑍𝑎 …................ (1.26) Where 𝜋∗ is the restricted profit (real value) from paddy production: total revenue less total cost of labour, seeds, fertilizer and agro chemicals normalized by the farm gate price of paddy; 𝑃𝐿 ∗ is the wage rate of labour per man day normalized by farm gate price of paddy; 𝑃𝑆 ∗ is the seed price normalized by farm gate price of paddy, 𝑃𝐹 ∗ is the fertilizer farm gate price normalized by farm gate price of paddy, 𝑃𝐶 ∗ is the agro chemical price normalized by farm gate price of paddy (import value of pesticides is included as a proxy for agro chemical price). Fixed inputs included in the specifications of the profit function are 𝑍𝑚 machinery price (obtained by dividing the machinery cost by GDP deflator index) and 𝑍𝑎 is the land area index. All the price indices are in real value (2000=1). The parameters 𝛼0, 𝛼, 𝛽, 𝛿, and 𝜃 are to be estimated and subscripts 𝐿, 𝑆, 𝐹, and 𝐶 stand for the variable input of production labour, seeds, fertilizer and agro chemicals respectively. The partial derivatives of normal restricted translog profit function (1.26) with respect to log of input price are the negative share equations for labour, seeds, fertilizer and agro chemicals as follows: − 𝑃𝐿𝑋𝐿 𝜋∗ = 𝛼𝐿 + 𝛾𝐿𝐿 ln𝑃𝐿 ∗ + γ LS ln𝑃𝑆 ∗ + 𝛾𝐿𝐹 ln𝑃𝐹 ∗ + γ LC ln𝑃𝐶 ∗ + 𝛿𝐿𝑍𝑚 ln𝑍𝑚 + 𝛿𝐿𝑍𝑎 ln𝑍𝑎…...... (1.27) − 𝑃𝑆𝑋𝑆 𝜋∗ = 𝛼𝑆 + 𝛾𝑆𝑆 ln𝑃𝑆 ∗ + γ LS ln𝑃𝐿 ∗ + 𝛾𝑆𝐹 ln𝑃𝐹 ∗ + 𝛾𝑆𝐶 ln𝑃𝐶 ∗ + 𝛿𝑆𝑍𝑚 ln𝑍𝑚 + 𝛿𝑆𝑍𝑎 ln𝑍𝑎 .......... (1.28) − 𝑃𝐹𝑋𝐹 𝜋∗ = 𝛼𝐹 + 𝛾𝐹𝐹 ln𝑃𝐹 ∗ + γ LF ln𝑃𝐿 ∗ + 𝛾𝑆𝐹 ln𝑃𝑆 ∗ + 𝛾𝐹𝐶 ln𝑃𝐶 ∗ + 𝛿𝐹𝑍𝑚 ln𝑍𝑚 + 𝛿𝐹𝑍𝑎 ln𝑍𝑎 ........ (1.29) − 𝑃𝐶𝑋𝐶 𝜋∗ = 𝛼𝐶 + 𝛾𝐶𝐶 ln𝑃𝐶 ∗ + γ LC ln𝑃𝐿 ∗ + 𝛾𝑆𝐶 ln𝑃𝑆 ∗ + 𝛾𝐹𝐶 ln𝑃𝐹 ∗ + 𝛿𝐶𝑍𝑚 ln𝑍𝑚 + 𝛿𝐶𝑍𝑎 ln𝑍𝑎 ........ (1.30) Where𝑋𝐿,𝑋𝑆 ,𝑋𝐹 and 𝑋𝐶 are the quantities of variable inputs of labour, seeds, fertilizer and agro chemicals, respectively. Other variables and parameters are same as defined earlier. Under the assumptions of profit maximizing and price taking behaviour, the parameters in equation (1.26) must be equal to the corresponding parameters in equation (1.27), (1.28), (1.29), (1.30) and must fulfil the symmetry restriction. This concept will provide testing the hypothesis of profit maximization. Since the input and output shares come from a singular system of equations (since by definition 𝑆𝑞 − 𝑆𝑖=1), one of the share equations, the output share is dropped and the profit and factor demand equations are estimated as simultaneous system. Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 28 An error term of the profit function and share equations are likely to be correlated contemporaneously due to large number of common explanatory variables. Thus Ordinary Least Square (OLS) is not applicable to estimate the equation in the system. OLS is also not appealing as we need to impose cross equation restrictions. This problem can be overcome by using Zellner’s estimation procedure for Seemingly Unrelated Regression (SUR) to obtain estimates which are asymptotically equivalent to Maximum Likelihood Estimation (MLE) when iterated to convergence and invariant to which share equation is deleted. In addition to the symmetry constraints (𝛾𝐿𝑆 = 𝛾𝑆𝐿 ,𝛾𝐿𝐹 = 𝛾𝐹𝐿 , 𝛾𝐿𝐶 = 𝛾𝐶𝐿 , 𝛾𝑆𝐹 = 𝛾𝐹𝑆 , 𝛾𝑆𝐶 = 𝛾𝐶𝑆 , 𝛾𝐶𝐹 = 𝛾𝐹𝐶 ), the linear parametric constraints are also imposed across equations. Before proceeding to the estimated parameter of normalized restricted translog profit and share equations, two hypothesis tests are carried out. They are test for the validity of profit maximization and Cobb Douglas hypothesis. The first empirical test checks the validity of symmetry and homogeneity restrictions across profit and share equations. The null hypothesis in the first test indicates that the parameters of the input share equations 1.27, 1.28, 1.29 and 1.30 are equal to the corresponding same parameters on the profit equation 1.26. An F test statistic with good asymptotic properties is conducted to test this hypothesis (Theil, 1971). F test statistics indicates that the null hypothesis cannot be rejected at 0.05 level of significance. This means that the profit maximization assumption is valid for all four districts. The second statistical test is conducted in order to check the Cobb-Douglas (C-D) hypothesis where the coefficients of all second order terms in profit function (1.26) should be zero. Therefore, an F- test is conducted to test the null hypothesis that all 𝛾𝑖ℎ equal zero and all 𝛿𝑖𝑘 equal zero. Based on the F-test, the hypothesis on Cobb-Douglas is rejected, suggesting that the translog profit function is more suitable for the data. The Stata 12 statistical software is used for the analysis. 3. RESULTS AND DISCUSSION The parameter estimates of translog profit function for each district is presented in Table 1 and output supply and demand elasticities derived are presented in Table 2, 3, 4 and 5 for the selected districts of Anuradhapura, Hambantota, Kurunegala, and Polonnaruwa respectively. These elasticities are evaluated at simple averages of the 𝑆𝑖 and at geometric means of the variable input prices and of levels of fixed inputs. Table 1: Estimated normalized translog profit function for paddy Variable Polonnaruwa Anuradhapura Hambantota Kurunegala Lnpl -0.810 * (0.114) -0.982 * (0.145) -0.955 * (0.129) -1.212 * (0.109) Lnps -0.086 * (0.010) -0.114 * (0.015) -0.174 * (0.023) -0.124 * (0.015) Lnpf -0.227 * (0.031) -0.244 * (0.029) -0.243 * (0.038) -0.279 * (0.037) Lnpc -0.125 * (0.024) -0.117 * (0.016) -0.171 * (0.034) -0.149 * (0.020) Lnplpl -0.745 * (0.143) -1.089 * (0.155) -0.941 * (0.141) -1.548 * (0.133) Lnpsps -0.060 * (0.007) -0.212 * (0.031) -0.024 (0.037) -0.092 * (0.019) Lnpfpf -0.118 * (0.014) -0.129 * (0.020) -0.168 * (0.026) -0.150 * (0.017) Lnpcpc 0.017 -0.073 0.179 * -0.082 Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 29 (0.029) (0.052) (0.089) (0.056) Lnplps -0.034 * (0.012) -0.027 (0.028) -0.167 * (0.044) -0.059 * (0.022) Lnplpf -0.093 * (0.035) -0.195 * (0.037) -0.199 * (0.046) -0.257 * (0.038) Lnplpc -0.202 * (0.042) -0.119 * (0.045) -0.372 * (0.085) -0.130 * (0.042) Lnpspf -0.009 * (0.003) -0.032 * (0.005) 0.001 (0.013) -0.018 * (0.006) Lnpspc -0.031 * (0.007) 0.025 (0.027) -0.102 * (0.041) -0.102 * (0.024) Lnpfpc 0.007 (0.011) 0.009 (0.010) -0.070 * (0.025) -0.022 * (0.009) Lnplzm -0.078 (0.207) -0.081 (0.150) 0.064 (0.073) 0.621 * (0.176) Lnplza -0.836 * (0.270) -0.747 * (0.153) -0.786 * (0.111) -1.538 * (0.212) Lnpszm -0.039 (0.012) -0.029 (0.022) 0.073 * (0.021) -0.023 (0.022) Lnpsza -0.043 * (0.019) -0.065 * (0.022) -0.242 * (0.029) -0.106 * (0.025) Lnpfzm -0.145 * (0.053) -0.024 (0.064) -0.151 * (0.045) -0.069 (0.058) lnpfza -0.284 * (0.082) -0.101 (0.069) -0.144 * (0.071) -0.198 * (0.064) Lnpczm 0.020 (0.059) -0.033 (0.039) -0.109 * (0.041) -0.021 (0.036) Lnpcza -0.175 (0.096) -0.012 (0.040) -0.060 (0.060) -0.132 * (0.041) Cons 7.658 (0.099) 7.303 * (0.162) 6.777 * (0.136) 7.207 * (0.102) Note: Standard errors are in parenthesis. * Significant at 5% As shown in Table 1, coefficient of factor prices of labour, seeds, fertilizer and agro chemicals and their interaction terms are negative in the profit function. Based on the results, price of labour, seeds, fertilizer and agro chemicals are significantly influence the profit function. Therefore, cheaper input prices increase the profit obtained from paddy cultivation. However, the coefficients values are less than unity, indicate the input prices affect inelastic on the profit. Moreover, the highest coefficient values for the labour wage rate, followed by fertilizer price indicate the profit is highly depending on labour wage rate and fertilizer price. As shown in Table 2, 3, 4 and 5, the farm gate price of paddy has a significant positive effect on paddy supply while variable factor prices have negative effects as expected. The supply elasticity with respect to previous year farm gate price is found to be positive and inelastic; a one percent rise in the paddy price would expand the supply of paddy ranging between 0.4 to 0.6 and on an average it is 0.5 percent. Therefore, results indicate that the effect of paddy price is more important to determine the country’s paddy supply than the price of fertilizer. And also our supply elasticity is in line with recent study findings of Weerahewa (2004) that estimated supply elasticity with respect to own price is 0.6 and with respect to fertilizer price is -0.07 for the 1978-2000 periods. But our elasticity values are smaller than the results obtained by Rajapaksha & Karunagoda (2009) that estimated own price elasticity of paddy supply raging between 0.85 in Polonnaruwa district to 2.37 in Kalutara district. Meantime, the aggregate paddy output elasticities with respect to price were relatively low in Sri Lanka in early years, with short-run supply elasticity falling between 0.09 and Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 30 0.13 and long-run elasticity 0.11-0.19 (Gunawardana & Oczkowski, 1992; Bogahawatte, 1983; Samaratunga, 1984). The negative supply elasticities with respect to variable inputs imply that the input use will decline with the increased input price thus would reduce the paddy yield and output. Moreover, the results show that wage rate and fertilizer price are the influencing variable inputs on paddy supply while fixed factor of land also affect significantly. The results show that the output supply with respect to fertilizer price is significant and -0.05 on an average. Therefore, fertilizer subsidy has a significant positive effect on paddy output in the country and the effect is relatively small compared to paddy price. According to the figures, supply elasticity with respect to fertilizer price is quite comparable in absolute values in Anuradhapura and Kurunegala districts (-0.05) whereas values are less than unity in all four districts. Thus, the paddy supply with respect to fertilizer price is inelastic as estimated. In addition, one percent rise in wage rate will decrease the paddy supply by 0.18 percent on an average. Moreover, expansion of cultivable area would enhance paddy supply significantly and positively in all the districts. The largest supply elasticity in Kurunegala with respect to land area may be due to the fact that farmers in Kurunegala district cultivate paddy using rain-fed as well as irrigated system and when the irrigated water is available adequately so that increasing in land area would expand the paddy production significantly. But in other three districts paddy cultivation largely depends on irrigated water and supply is limited. Moreover, fixed input of machinery has slight positive impact in all the districts except in Kurunegala. The own price elasticity of demand for inputs have negative signs as expected and are statistically significant except own price elasticity of agro chemical in Anuradhapura andprice elasticity of seed and chemical in Kurunegala district. Elasticity values are less than unity in absolute value for all the inputs in all four districts (except agro chemical in Hambantota and Polonnaruwa). Derived elasticity estimates for paddy supply and demand for variable inputs in rice production Table 2: Elasticity values – Anuradhapura district Output/input Price of Paddy Price of Labour Price of Seed Price of Fertilizer Price of Chemical Machinery Land Paddy supply 0.478 * -0.231 * -0.001 -0.054 ** -0.042 0.089 0.503 * Labour demand 0.617 * -0.588 * -0.078 * 0.022 0.027 0.044 0.404 * Seed demand 0.031 -0.609 * -0.830 * 0.087 -0.339 0.013 0.028 Fertilizer demand 0.596 ** 0.090 0.047 -0.578 * -0.155 * .015 0.065 Agro-chemical demand 0.868 0.210 -0.334 -0.286 * -0.457 0.016 0.006 Note: The figures are calculated at the mean value of shares. Significance test for each estimated elasticity can be conducted by making use of estimated standard errors as follows: 𝑆𝐸 𝜂𝑖ℎ = 𝑆𝐸(𝛾𝑖ℎ)/𝑆𝑖 and t statistics can be calculated by 𝑡𝑖ℎ = 𝜂𝑖ℎ/𝑆𝐸(𝜂𝑖ℎ) where 𝜂𝑖ℎ is elasticity of factor demand * Significant at 5%, ** Significant at 10% Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 31 Table 3: Elasticity values – Hambantota district Output/input Price of Paddy Price of Labour Price of Seed Price of Fertilizer Price of Chemical Machinery Land Paddy supply 0.422 ** -0.019 -0.002 -0.009 0.012 0.095 0.667 * Labour demand 0.056 -0.535 * 0.086 0.057 0.336 * -0.034 0.378 * Seed demand 0.040 0.499 -0.948 * -0.215 * 0.623 * -0.035 0.116 Fertilizer demand 0.100 -0.505 * -0.140 -0.374 * 0.198 0.109 0.104 Agro-chemical demand 0.521 1.708 * 0.547 * 0.267 -2.339 * 0.055 0.030 Note: The figures are calculated at the mean value of shares. Significance test for each estimated elasticity can be conducted by making use of estimated standard errors as follows: 𝑆𝐸 𝜂𝑖ℎ = 𝑆𝐸(𝛾𝑖ℎ)/𝑆𝑖 and t statistics can be calculated by 𝑡𝑖ℎ = 𝜂𝑖ℎ/𝑆𝐸(𝜂𝑖ℎ) where 𝜂𝑖ℎ is elasticity of factor demand * Significant at 5%, ** Significant at 10% Table 4: Elasticity values –Kurunegaladistrict Output/input Price of Paddy Price of Labour Price of Seed Price of Fertilizer Price of Chemical Machinery Land Paddy supply 0.462 * -0.168 ** -0.001 -0.060 * 0.006 -0.298 * 1.205 * Labour demand 0.422 ** -0.396 * -0.051 * 0.021 0.003 -0.374 * 0.928 * Seed demand 0.024 -0.444 * -0.285 -0.079 0.785 * 0.012 0.054 Fertilizer demand 0.607 * 0.086 -0.037 -0.618 * -0.038 0.052 0.148 Agro-chemical demand 0.120 0.022 0.670 * -0.070 -0.501 0.012 0.076 Note: The figures are calculated at the mean value of shares. Significance test for each estimated elasticity can be conducted by making use of estimated standard errors as follows: 𝑆𝐸 𝜂𝑖ℎ = 𝑆𝐸(𝛾𝑖ℎ)/𝑆𝑖 and t statistics can be calculated by 𝑡𝑖ℎ = 𝜂𝑖ℎ/𝑆𝐸(𝜂𝑖ℎ) where 𝜂𝑖ℎ is elasticity of factor demand * Significant at 5%, ** Significant at 10% Table 5: Elasticity values-Polonnaruwa District Output/input Price of Paddy Price of Labour Price of Seed Price of Fertilizer Price of Chemical Machinery Land Paddy supply 0.653 * -0.281 * -0.033 -0.113 * -0.052 0.085 0.971 * Labour demand 0.795 ** -0.837 * -0.003 -0.101 * 0.145 -0.053 0.676 ** Seed demand 0.827 -0.022 -0.439 * -0.062 -0.304 0.012 ** 0.036 Fertilizer demand 1.143 * -0.361 * -0.025 -0.679 * -0.079 0.109 0.185 Agro-chemical demand 0.905 0.897 -0.214 -0.136 -1.452 ** 0.016 0.074 Note: The figures are calculated at the mean value of shares. Significance test for each estimated elasticity can be conducted by making use of estimated standard errors as follows: 𝑆𝐸 𝜂𝑖ℎ = 𝑆𝐸(𝛾𝑖ℎ)/𝑆𝑖 and t statistics can be calculated by 𝑡𝑖ℎ = 𝜂𝑖ℎ/𝑆𝐸(𝜂𝑖ℎ) where 𝜂𝑖ℎ is elasticity of factor demand * Significant at 5%, ** Significant at 10% Increase in output price would also encourage direct and significant expansion in demand for variable inputs especially labour and fertilizer in paddy cultivation. In quantitative terms, the percent increases in demand for labour, associated with one percent increase in output price are 0.62, 0.06, 0.42 and 0.80 for the Anuradhapura, Hambantota, Kurunegala and Polonnaruwa districts, respectively. Meantime, the results also indicate that labour and agro chemicals in all four districts are substitute inputs in paddy production. Elasticity of labour with respect to fixed inputs of land is positive and significant for all the districts and the average elasticity value is 0.60. Therefore, degree of responsiveness to labour absorption is significant but inelastic. The significant negative elasticity value of labour demand with respect to machinery in Kurunegala district indicates that labour and machinery are complementary inputs and their combined application will increase paddy production. This might be due to the fact that in Sri Lanka the majority of the paddy lands are small plots of less than one hectare, thus mechanization in large scale cannot apply and consequently labour and small equipment are jointly use in paddy cultivation. Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 32 The study result suggests that the demand for seeds with respect to labour and fertilizer are complementary in inputs. Meanwhile, seed demand increase slightly with fixed inputs of machinery and land. Demand for fertilizer with respect to labour is negative and significant in the Hambantota and Polonnaruwa districts, while the elasticity values are positive but insignificant in Anuradhapura and Kurunegala districts. Hence, fertilizer and labour inputs in Hambantota and Polonnaruwa districts act as complementary inputs. Fertilizer demand elasticity with respect to land is in around 0.1. As can be seen from Tables 2, 3, 4 and 5, the own price elasticities of fertilizer are -0.58, -0.37, -0.62 and -0.68 for Anuradhapura, Hambantota, Kurunegala and Polonnruwa districts respectively. Furthermore, fertilizer demand elasticity to the output price is highest in Polonnaruwa (1.14). Contrary, one percent increase in paddy price will increase the fertilizer demand only by 0.10 percent in Hambantota district. This is due to the fact that in Polonnaruwa district fertilizer use in paddy farming is noticeably higher than that of Hambantota district. Therefore, findings of our results show that fertilizer demand in the country is significantly affected by both farm gate price of paddy and fertilizer price. However, Ekanayake (2006) using a simple regression model, found that changes in the prices of fertilizer do not have a significant effect on fertilizer usage and own price elasticity of fertilizer demand is -0.1. Rajapaksha & Karunagoda (2009) also concluded that the fertilizer demand in the country is highly responsive to the paddy price rather than fertilizer price. According to their estimation, fertilizer demand is elastic with respect to its own price and paddy price. However, normalized translog profit function estimated by them seems to be unclear since they used interaction between paddy price and variable inputs in their model. Based on simple Cobb Douglas approach using individual farm level data, Weligamage et al. (2009) also found that the elasticity of paddy yield with respect to fertilizer application is 0.15 and this estimate implies that the elasticity of fertilizer demand with respect to fertilizer price is approximately -1.2. Study results indicate that agro chemical demand with respect to output price is positive but not significant. Similarly, agro chemical demand with respect to fixed inputs is positive and relatively small. Moreover, the effect of wage rate on agro chemical demand is positive in all the districts indicates labour and agro chemicals are substitutes in paddy cultivation. This might be because instead of using more labour in general land preparation activities, weedicides can be used intensively to reduce the labour inputs. In addition, seed is also substitute to agro chemicals in Hambantota and Kurunegala districts. The analysis also suggests that the expansion of capital in the form of agricultural machineries, decreases labour demand (except in Anuradhapura district) contributes positively to paddy production. In addition, exogenous increases in land area also increase the paddy production and demand for all variable inputs of production in the country. The expansion of land has relatively large influence on output supply and labour demand while seed, fertilizer and agro chemical demand have small impact. The estimated elasticities of substitution between pairs of inputs are presented in the Table 6 for four districts. Labour is a substitute for agro chemicals and the highest substitutability value can be observed in Hambantota district. Nonetheless, labour is complement to the seed except in Hambantota district. The estimates also indicate that seed and fertilizer are complementary except in Anuradhapura district. In addition elasticity of substitution values between fertilizer and agro chemicals are also negative (except Hambantota). As seen from Table 6, the largest substitutability relation is between seeds and agro chemicals and the value is higher than the unity. However, the low elasticity of substitution between labour and fertilizer and other input pairs indicates that there is a complementary relationship among these inputs hence combined application of such inputs will synergistically increase the paddy production in the country. Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 33 Table 6: Estimated elasticities of substitution Anuradhapura 𝜎𝐿𝐿=-0.69 𝜎𝐿𝑠= -0.71 𝜎𝐿𝐹=0.11 𝜎𝐿𝐶=0.24 𝜎𝑠𝑠=-7.56 𝜎𝑆𝐹=0.43 𝜎𝑆𝐶=-3.05 𝜎𝐹𝐹=-2.81 𝜎𝐹𝐶=-1.39 𝜎𝐶𝐶=-4.1 Hambantota 𝜎𝐿𝐿=-0.70 𝜎𝐿𝑠=0.65 𝜎𝐿𝐹=0.28 𝜎𝐿𝐶=2.23 𝜎𝑠𝑠=-7.17 𝜎𝑆𝐹=-1.06 𝜎𝑆𝐶=4.14 𝜎𝐹𝐹=-1.84 𝜎𝐹𝐶=1.31 𝜎𝐶𝐶=-15.53 Kurunegala 𝜎𝐿𝐿=-0.40 𝜎𝐿𝑠=-0.45 𝜎𝐿𝐹=0.09 𝜎𝐿𝐶=0.02 𝜎𝑠𝑠=-2.55 𝜎𝑆𝐹=-0.33 𝜎𝑆𝐶=5.99 𝜎𝐹𝐹=-2.56 𝜎𝐹𝐶=-0.29 𝜎𝐶𝐶=-3.83 Polonnaruwa 𝜎𝐿𝐿=-1.06 𝜎𝐿𝑠=-0.03 𝜎𝐿𝐹=-0.46 𝜎𝐿𝐶=1.14 𝜎𝑠𝑠=-4.91 𝜎𝑆𝐹=-0.28 𝜎𝑆𝐶=-2.39 𝜎𝐹𝐹=-3.09 𝜎𝐹𝐶=-0.62 𝜎𝐶𝐶=-11.41 Note: Elasticities of substitution values are evaluated at the means of the data. The subscripts L, S, F and C stand for Labour, Seeds, Fertilizer and Agro chemicals respectively. The own elasticities of substitution has little economic meaning. From the analysis, following broad conclusions may be drawn about the paddy production in Sri Lanka. 4. CONCLUSION The overall estimations suggest that the changes in market prices of inputs and output affect significantly on farmer profit, rice supply and the resource use in paddy cultivation. Meanwhile, farmers maximize profit in paddy farming subject to given inputs (labour, seeds, fertilizer and agro chemicals), paddy prices and fixed factors of production (machinery and land). The results show that paddy output in the country is more responsive to the increase in output price than change in fertilizer price. The impact of fertilizer price on paddy supply is relatively small (supply elasticity with respect to fertilizer price is -0.05) but significant.Generally fertilizer demand is thought to be more elastic in developing countries; this study outcome confirms that in Sri Lanka fertilizer demand is inelastic but significant to its own price. Consequently, fertilizer subsidy is one of the main factors to increase the fertilizer demand as well as output supply in the country. As indicated by the elasticity, fertilizer demand in the country is also significantly depend on the output price of supply. Therefore, paddy price has strong incentive to promote the fertilizer demand as well as paddy production in the country. In addition, the analysis shows that labour demand in paddy cultivation is more responsive to the wage rate while fertilizer price has fair degree of responsiveness. The low substitutability between labour and fertilizer and other input pairs indicates combined application of such inputs increase the paddy production in the country.Overall, this study suggests that farmers are price sensitive and assures prevalence of high output price is essential for higher production. Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 34 Views and opinions expressed in this study are the views and opinions of the authors, Asian Journal of Agriculture and Rural Development shall not be responsible or answerable for any loss, damage or liability etc. caused in relation to/arising out of the use of the content. References Altemeier, K., & Bottema, T. (1991). Agricultural diversification in Indonesia: Price responses and linkages in the food crop sector, 1969-1998; An Outlook to 2000. Working Paper 11, The CGPRT Center, Indonesia. Atkinson, S. E., & Halvorsen, R. (1976). Interfuelsubstitution in electric power generation.Journal of Political Economy, 84(5), 959-978. Ball, V. E. (1988). Modelingsupply response in a multiproduct framework. American Journal of Agricultural Economics, 70(4), 813-825. Bapna, S. L., Binswanger, H. P., & Quizon, J. B. (1984). System of output supply and factor demand equations for semi-arid tropical India. Indian Journal of Agricultural Economics, 39, 179-202. Beccera, P. A. V., & Shuway, C. R. (1992). Multiple output production modeled with three functional form. Journal of Agricultural and Resource Economics, 17(1), 13-28. Berndt, E. R., & Christensen, L. R. (1973). The translog function and substitution of equipment, structures and labour in U.S. manufacturing 1928-68. Journal of Econometrics, 1, 81-113. Berndt, E.R. &Wood, D. O. (1975). Technology, prices and derived demand for energy. Review of Economics and Statistics, 52, 259-268. Binswanger, H. P. (1974). A cost function approach to the measurement of elasticities of factor demand and elasticities of substitution. American Journal of Agricultural Economics, 56, 377-386. Bogahawatte, C. (1983). Simulating the impact of changes in wheat flour imports and world price of rice on the supply and demand for rice in Sri Lanka. Indian Journal of Agricultural Economics,38, 15-26. Central Bank of Sri Lanka (2014). Annual Report. Colombo, Sri Lanka. Chaudhary, M. A., Khan, M. A., & Naqvi, K. H. (1998). Estimates of farm output supply and input demand elasticities: The translog profit function approach. The Pakistan Review, 37(4), 1031-1050. Christensen, L. R., Jorgenson, D. W., & Lau, L. J. (1973). Transcendental logarithmic production frontiers.The Review of Economics and Statistics, 55(1), 28-45. Christensen, R., Jorgenson, D. W., & Lau, L. R. (1975). Transcendental logarithmic utility functions. American Economic Review, 65, 367-833. Department of Agriculture (1990-2012). Cost of cultivation of agricultural crops. Socio Economic and Planning Centre, Peradeniya, Sri Lanka. Department of Census and Statistics (2002). Census of Agriculture, Colombo, Sri Lanka.Accessed July 2015 fromhttp://www.statistics.gov.lk/. Department of Census and Statistics (2012).Paddy Statistics. Colombo, Sri Lanka. Accessed July 2015 from http://www.statistics.gov.lk/. Diewert, W. E. (1971). An application of the shephard duality theorem: A generalized Leontief production function. Journal of Political Economics, 79, 481-507. Diewert, W. E. (1973). Functional forms for Profit and Transformation Functions. Journal of Economic Theory, 6, 284-316. Diewert, W. E. (1974). Application of duality theory. In Frontiers of Quantitative Economics, vol. 2, edited by M.D. Intrilligator and David Kendrick. Amsterdam: North Holland Publishing Co. Ekanayake, H. K. J. (2006). Impact of fertilizer subsidy in paddy cultivation in Sri Lanka. Staff Studies, Central Bank of Sri Lanka, 36(1 & 2), 73-92. Fulginiti, E. L., & Perrin, R. K. (1990). Argentine agricultural policy in a multiple-input, multiple output frame work. American Journal of Agricultural Economics, 72(2), 279-288. http://www.statistics.gov.lk/ http://www.statistics.gov.lk/ Asian Journal of Agriculture and Rural Development, 6(2)2016, 21-35 35 Fuss, M. D., McFadden, D., & Mundlak, Y. (1978). A survey of functional forms in the economic analysis of production. In Production Economics: A Dual Approach to Theory and Applications, edited by Fuss, M. and McFadden, D. Amsterdam: North-Holland Publishing Co. Gunawardana, P. J., & Oczkowski, E. A. (1992). Government policies and agricultural supply response: Paddy in Sri Lanka. Journal of Agricultural Economics, 43(2), 231-242. Hattink, W., Heerink. N., & Thijssen, G. (1998). Supply Response of Cocoa in Ghana: A Farm- level Profit Function Analysis. Journal of African Economies, 7(3), 424-444. Lopez, R. (1985). Structural implications of a class of flexible functional forms for profit functions. International Economic Review, 26, 593-601. Rajapaksha, R. D. D. P., & Karunagoda, K. S. (2009). Fertilizer demand for paddy cultivation in Sri Lanka with special reference to fertilizer subsidy program. Sri Lanka Journal of Agrarian Studies, 13(2), 25-38. Ray, S. C. (1982). A translog cost function analysis of U.S. Agriculture, 1939-77. American Journal of Agricultural Economics, 64(3), 490-498. Samaratunga, P. A. (1984). An aggregate economic analysis of the effects of the policy on price and distribution of rice in Sri Lanka. University of the Philippines, Los Banos, unpublished MSc thesis. Sidhu, S. S., & Baanante, C. A. (1979). Farm-level fertilizer demand for Mexican wheat varieties in the Indian Punjab. American Journal of Agricultural Economics, 61(3), 455-462. Siregar, H. (2007). Elasticities of output supply and input demand of Indonesian food crops and their policy implication: multi-input multi output framework.Socio Economic of Agriculture and Agribusiness, 7(2), 1-15. Theil, H. (1971). Principles of Econometrics, New York: John Wiley and Sons. Thiruchelvum, S. (2005). Efficiency of rice production and issues relating to cost of production in the districts of Anuradhapura and Polonnaruwa. Journal of National Science Foundation, Sri Lanka, 33(4), 247-256. Thusiman, J. A., Gunawardena, P., & Flinn C. J. (1987). Supply response and fertilizer demand in Sri Lanka.Quarterly Journal of International Agriculture, 26(4), 355-367. Wall, C. A., & Fisher, B. (1988). Supply Response and the Theory of Production and Profit Functions. Review of Marketing and Agricultural Economics, 56(3), 383-404. Weerahewa, J. (2004). Impacts of trade liberalization and market reforms on the paddy/rice sector in Sri Lanka. MTID Discussion Paper No. 70, Washington D.C.: International Food Policy Research Institute (IFPRI). Accessed June 2014 from http://www.ifpri.org/publication/impacts-trade-liberalization-and-market-reforms-rice- sector-sri-lanka/. Weerasooriya, S., & Gunaratne, L. H. P. (2009). Impact of fertilizer subsidy on productivity, supply and cultivated extent of paddy. Abstracts of the Third Annual research Forum of Sri Lanka Agricultural Economics Association, October 2, Hector Kobbekaduwa Agrarian Research and Training Institute, Colombo. Accessed February 2015 from http://www.slageconr.net/saea/arf/abstractsarf03.pdf/. Weligamage, P. C. R., Shumway, K. A., Blatner, W. R., Butcher, & Giordano, M. (2009). Production Function for Rice in Kirindi Oya Major Irrigation System in Sri Lanka. Presentation at the Third Annual Research Forum of Sri Lanka Agricultural Economic Association, October 2, Hector Kobbekaduwa Agrarian Research and Training Institute, Colombo. World Bank (2013). What is the cost of a bowl of rice? The impact of Sri Lanka’s current trade and price policies on the incentive framework for agriculture. Discussion Paper 02, Washington, DC 20433. Yotopoulos, P. A., Lau, L. J., & Lin, W. (1976). Microeconomic output supply and factor demand functions in the agriculture of the province of Taiwan. American Journal of Agricultural Economics, 58(2), 333-340. http://www.ifpri.org/publication/impacts-trade-liberalization-and-market-reforms-rice-sector-sri-lanka/ http://www.ifpri.org/publication/impacts-trade-liberalization-and-market-reforms-rice-sector-sri-lanka/ http://www.slageconr.net/saea/arf/abstractsarf03.pdf/