Efficient Technology and the Conservation of Natural Forests: Evidence from Sri Lanka

Gunatilake, Herath Working Paper Efficient Technology and the Conservation of Natural Forests: Evidence from Sri Lanka ERD Working Paper Series, No. 105 Provided in Cooperation with: Asian Development Bank (ADB), Manila Suggested Citation: Gunatilake, Herath (2007) : Efficient Technology and the Conservation of Natural Forests: Evidence from Sri Lanka, ERD Working Paper Series, No. 105, Asian Development Bank (ADB), Manila, https://hdl.handle.net/11540/1863 This Version is available at: https://hdl.handle.net/10419/109306 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 Printed in the Philippines Efficient technology and the conservation of natural Forests: Evidence from sri Lanka Herath Gunatilake October 2007 about the paper Herath Gunatilake examines the feasibility of technical efficiency improvement as an approach for forest conservation using a case study of saw milling in Sri Lanka. The paper shows that reduction of existing inefficiency helps prevent deforestation. Having compared the merits of this approach with other policy options, the paper asserts that technological improvements as a means of forest conservation deserve the attention of policymakers. Asian Development Bank 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org/economics ISSN: 1655-5252 Publication Stock No. about the asian development Bank ADB aims to improve the welfare of the people in the Asia and Pacific region, particularly the nearly 1.9 billion who live on less than $2 a day. Despite many success stories, the region remains home to two thirds of the world’s poor. ADB is a multilateral development finance institution owned by 67 members, 48 from the region and 19 from other parts of the globe. ADB’s vision is a region free of poverty. Its mission is to help its developing member countries reduce poverty and improve their quality of life. ADB’s main instruments for helping its developing member countries are policy dialogue, loans, equity investments, guarantees, grants, and technical assistance. ADB’s annual lending volume is typically about $6 billion, with technical assistance usually totaling about $180 million a year. ADB’s headquarters is in Manila. It has 26 offices around the world and more than 2,000 employees from over 50 countries. ERD WoRking PaPER SERiES no. 105 ERD Working Paper No. 105 EfficiEnt tEchnology and thE consErvation of natural forEsts: EvidEncE from sri lanka hErath gunatilakE octobEr 2007 Herath Gunatilake is a Senior Economist at the Economic Analysis and Operations Support Division, Economics and Research Department, Asian Development Bank. This research was carried out with financial assistance from the Economy and Environment Program for South East Asia (EEPSEA). The author acknowledges comments received from Nancy Olewiler, Department of Economics, Simon Fraser University during the design and field research phases of this work. The author also acknowledges valuable comments by William F. Hyde, Senior Scientist, Forest Economics and Policy Analysis Research Center, University of British Columbia, who helped in reorganizing the original version of the paper. The author is, however, solely responsible for any errors contained in the paper. Asian Development Bank 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org/economics ©2007 by Asian Development Bank October 2007 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. 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. CoNtENts Abstract vii I. INTROD�CTIONI. INTROD�CTION 1 II. BAC��RO�NDII. BAC��RO�ND 2 III.III. METhODOlO�y fOR CAlC�lATIN� TEChNICAl EffICIENCy 5 I�. SAMPlE, DATA, AND hyPOThESESI�. SAMPlE, DATA, AND hyPOThESES 7 �. RES�lTS 1�. RES�lTS 10 A. Ef��ciency Improvements and forest Conservation 1A. Ef��ciency Improvements and forest Conservation 12 B. Ef��ciency Improvement versus Other Policies 14 �I. CONCl�DIN� COMMENTS 1�I. CONCl�DIN� COMMENTS 16 Appendix�� Distribution of the Sample and Descriptive Statistics of the Data 1Appendix�� Distribution of the Sample and Descriptive Statistics of the Data 18 References 2References 20 sectiOn iii methOdOlOgy fOr calculating technical efficiency erd WOrking PaPer series nO. 105 5 sawmills2 is about 750m3/year. Sawmills in Sri lanka are old, with the average age of surveyed mills at about 18 years (MfE 1995). The oldest mills were established in the early 1950s and many have not gone through any major improvements. The old mills employ simple and labor-intensive technology and still use originally imported equipment, while some of the new mills have locally made replicas of the old imported machinery (MfE 1995). Primary cutting machines are dominated by circular saws, and some larger mills have band saws (with horizontal reciprocating head rings) that allow for higher recovery of wood due to their relatively narrow saw blades. Small size, labor intensiveness, old and mostly worn-out machinery, poor layout, and poor saw-doctoring and feeding systems have resulted in heavy losses during milling. The average recovery rate is only about 40% (compared to 55% and 50% in Malaysia and Indonesia, respectively). This picture of the sawmilling industry in Sri lanka implies an ongoing wastage and lack of ef��ciency. There may be possibilities for improving ef��ciency in this sector to make sawmilling more pro��table. Consumers may be able to purchase sawn timber at a lower price if ef��ciency is improved. More importantly, improving the ef��ciency of this sector may play a vital role in the conservation of natural forests. Reduction of wastage in sawmilling would relieve extra pressure on tree harvesting and reduce the ongoing destruction of natural forests in Sri lanka. This will have a signi��cant positive impact on forest conservation in Sri lanka, in addition to direct producer and consumer bene��ts. III. MEthoDoloGy FoR CAlCUlAtING tEChNICAl EFFICIENCy Traditionally, economic ef��ciency at the ��rm level is measured by single factor productivity. This approach is, however, not very accurate as other factors should be held constant in measuring factor productivity. farrell (1957) developed better and simple two measures of ef��ciency�� technical ef��ciency and allocative ef��ciency. Of these, technical ef��ciency reflects the ability of a ��rm to obtain maximum output from a given set of inputs, or to obtain a given level of output from a minimum level of inputs. Allocative ef��ciency reflects the ability of a ��rm to use inputs in optimal proportions, given their prices (Coelli 1995). These two ef��ciency measures are combined to estimate economic ef��ciency. farrell’s technical, allocative, and economic ef��ciency can be further elaborated on using the concept of unit isoquants (figure 2). Consider a ��rm producing output y from inputs X1 and X2 with the production function, y = f (X1, X2). Assuming constant return to scale, the frontier technology can be represented by the unit isoquant, 1 = f (X1/y, X2/y), QQ’. let WW’ represent the ratio of input prices. farrell de��nes a ��rm producing at point A as technically inef��cient and the ratio OB/OA gives farrell’s measure of technical ef��ciency. If the ��rm is operating at point B, it is considered technically ef��cient but allocatively inef��cient; the ratio OD/OB gives farrell’s measure of allocative ef��ciency. finally, the ratio OD/OA measures total ef��ciency. Economic ef��ciency is measured by combining both ratios and is equal to the product of technical and allocative ef��ciencies. 2 In Thailand, the average is 7,000mIn Thailand, the average is 7,000m3/year and in Indonesia it is 30,000 m3/year. Thus, the average output of a sawmill in Sri lanka is lower than that of other Asian countries. 6 OctOber 2007 efficient technOlOgy and the cOnservatiOn Of natural fOrests: evidence frOm sri lanka herath gunatilake FIGURE 2 EFFICIENCY MEASUREMENTS X2/Y Q C B Q’ A D OX1/Y Source: Gunatilake and Gunaratne (2002). W W’ Often, ��rm level input–output relationships are examined with production functions estimated using regression analysis. Since the regression line is ��tted through the means of the data set, such analysis provides only an average relationship (Alauddin et al. 1993). In contrast, the frontier production function corresponds to the formal de��nition of a production function, which refers to the maximum output obtainable from a given set of inputs and technology. The basic difference of a stochastic frontier, compared to an average production function, lies in the formulation of the residual term of the regression equation. here the error term is separated into symmetric and asymmetric components. The symmetric component represents the usual random variations, measurement errors, and statistical noise. The asymmetric (one-sided) term captures the technical inef��ciency of the ��rm (�umbhakar et al. 1991, Bravo-�reta and Pinheiro 1993, Coelli 1995). The major weaknesses of the stochastic frontier method include arbitrary speci��cation of the distributional form of the one-sided error term, selection of the functional form, and dif��culties involved when multiple outputs are present.3 following the standard assumption of Zellner et al. (1966), the above theory can be used to specify a frontier production function for sawmills. Assuming that mill owners maximize expected pro��ts, the single equation Cobb-Douglas stochastic production model (Aigner et al. 1977, Meeusen and van den Broeck 1977) can be speci��ed as�� Lny Ln Lnx v u i k ik ik i i = + + − = ∑β 0 1 5 β where yi is the index of sawn wood output, cubic feet/month x1 is the index of log inputs, cubic feet/month x2 is units of energy used, kilowatt hours/month 3 Technical ef��ciency of a multiple product ��rm can be measured using the mathematical programming method knownTechnical ef��ciency of a multiple product ��rm can be measured using the mathematical programming method known as data envelopment analysis. sectiOn iv samPle, data, and hyPOtheses erd WOrking PaPer series nO. 105 7 x3 is capital expenditure of the mill x4 is skilled labor, person days/month x5 is unskilled labor, person days/month βk (k =1,2,…5) are the parameters vi is a random variable, iid ∼N(0, σ 2 v ) ui is a non-negative random variable that represents technical inef��ciency The two error terms make the difference between an average production function and a frontier production function. �sing Battese and Coelli’s (1992) parametric speci��cation, the maximum likelihood estimation of the equation provides estimators for β, σ2 = σ2v + σ2u and γ = σ2u + σ2. Prediction of technical ef��ciency of a ��rm is based on conditional expectation of ui (exp(-ui)), given the value of random variable εi = vi - ui. Subtracting exp(-vi) from both sides of the above equation�� Lnyi* = Lnyi - ui where yi* = is the ith ��rm’s observed output, adjusted for statistical noise. The above equation forms the basis for the measurement of the technical ef��ciency of the ��rm. IV. sAMPlE, DAtA, AND hyPothEsEs Primary data for estimating the above production function was collected using a structured questionnaire. A strati��ed random sample was drawn from the population, based on the geographic zones in Sri lanka. from each zone, a number of mills to be included in the sample were decided based on the proportion of the mills in the zone. Then random number tables were used to select the predetermined number of mills. The original sampling framework and the distribution of the ��nal sample are given in the Appendix (Table A1). The survey was conducted in two stages. In the ��rst stage of the survey, information on general aspects of the sawmills was collected and an input–output sheet was given to the mill manager. At the ��rst stage, most of the managers agreed to ��ll out the input–output sheet. however, at the time of collection, almost all managers did not ��ll out the input–output sheets. With this disappointing experience, a second visit to the mills was arranged to ��ll the forms using person-to-person interviews. The data gathered are based on the memory of the mill manager in most cases, as they do not keep proper records. Data would have been more accurate had the mill managers ��lled out the form as inputs came in and soon after milling the logs. In the original sampling framework, the sample size was 180 mills. Thirteen mills were allocated for Ampara and Batticaloa districts. Due to security reasons, the research assistants were not able to make the second visit to these 13 mills. About eight questionnaires were excluded from the sample from different districts due to inconsistencies found in the answers. Another 11 mills were excluded from the sample due to mill owners’ reluctance to provide information on log inputs and outputs. Thus, a complete set of data was available for only 148 sawmills. Another problem encountered in estimating the equation was the presence of a number of different outputs. As mentioned earlier, the stochastic frontier technique can be used only for 8 OctOber 2007 efficient technOlOgy and the cOnservatiOn Of natural fOrests: evidence frOm sri lanka herath gunatilake single-output ��rms. Therefore, the different outputs were aggregated to a single output index using the following formula�� y p q Pn j rj rj r s j j n == = ∑ ∑ 1 1 _ where yj is the normalized output for the jth ��rm, s denotes the number of differentiated products, prj denotes the price of the rth product for the jth ��rm, qrj denotes the amount of rth product for the jth ��rm and, n is the number of ��rms. The average price in the denominator is de��ned as�� p p q q q qjrj rj r s j j rj r s − = = = = ∑ ∑ 1 1 / , A similar problem was encountered in measuring log inputs. The log inputs were broadly categorized into softwood and hardwood, disregarding the species, and were also aggregated using the above formula. The number of units of energy used during the month under consideration was obtained from the monthly electricity bills. An attempt was made to get accurate information on capital expenditure of the sawmills. however, during the pre-testing stage it was felt that mill owners/managers were not willing to reveal true information on capital expenditures, probably due to tax evasive strategies. Therefore, a proxy—mill capacity—was used in place of capital expenditure. Many technical ef��ciency studies have estimated a second regression equation4 to identify the determinants of technical ef��ciency. The study followed this approach to identify the determinants of technical ef��ciency of sawmilling. Since there are no previous studies on the technical ef��ciency of sawmilling, the variables were identi��ed based on the survey ��ndings, informal discussions with the mill managers, and scatter plots of ef��ciency scores and related variables. The variables and the postulated hypotheses are given in Table 1. 4 The ��rst regression equation estimates the production function to calculate ef��ciency scores. sectiOn iv samPle, data, and hyPOtheses erd WOrking PaPer series nO. 105 9 tablE 1 dEtErminants of tEchnical EfficiEncy variablE dEscription hypothEsizEd rElationship to tEchnical EfficiEncy X1 Age of the manager/owner Positive X2 Quality of log inputs Positive X3 Charges based on log input =1, Others = 0 Negative X4 Owner-managed mills = 1, Others = 0 Positive X5 Education of the manager/owner Positive X6 Entrepreneurship Positive X7 Capacity of the mill Negative X8 Source of energy, only public electricity =1, public electricity and other = 0 Negative X9 Age of the machines Negative As to age of the manager/owner, age is assumed to influence technical ef��ciency positively, since age reflects the experience of the manager in the sawmilling industry. Regarding quality of log inputs, given the scarcity of wood, many immature trees are harvested, resulting in very low recovery rates. Moreover, since the logs are not supplied by well-managed forestlands, many logs are not straight. high-quality logs (mature straight logs) are assumed to provide higher technical ef��ciency. for charges based on log input, certain mills only lease their machines for milling, with charges based on log inputs. The other mills purchase logs, mill them, and sell the sawn wood. Since the former type of mills have no incentives to improve technical ef��ciency, it was assumed that such mills are technically inef��cient. for the variable mill ownership, the survey revealed that when hired managers manage the mills, they are paid a ��xed monthly salary. There are no incentive payments based on pro��ts or any other measure of performance (�unatilake and �unaratne 2002). Therefore, owner-managed mills are assumed to be more ef��cient. for the issue on education of the manager/owner, formal education may enhance the management ability of managers. Therefore, it is assumed that a higher level of formal education positively affects technical ef��ciency. Meanwhile, on the variable entrepreneurship, managers with higher entrepreneurship ability may consciously take steps to increase technical ef��ciency to increase pro��ts. Therefore entrepreneurship is assumed to affect technical ef��ciency positively. On mill capacity, given log shortages, large mills may not be able to fully utilize their ��xed production factors. Therefore, it is assumed that mill capacity negatively affects technical ef��ciency. As to source of energy, Sri lanka has a very irregular supply of power because the country depends mainly on hydropower. Power cuts are frequent and, as mill owners explain, these power cuts may be one reason for lower ef��ciency. Therefore, it is assumed that mills with only public electricity supply are inef��cient compared to those with public as well as their own means of power supply (generators). finally, on age of machines, old and worn-out machines lead to low recovery and are therefore assumed to affect technical ef��ciency negatively. 10 OctOber 2007 efficient technOlOgy and the cOnservatiOn Of natural fOrests: evidence frOm sri lanka herath gunatilake The age of the owner (X1) was considered when the owner himself managed the mill. Otherwise, the manager’s age was considered. log quality (X2) was ranked from 1 to 5; 1 representing the poorest quality and 5 representing the best quality. The type of milling (X3) variable was measured as a dummy�� 1 for the mills that leased their machines, 0 otherwise. Owner-managed mills (X4) were assigned 1 while the mills managed by hired managers were given 0. Education (X5) was ranked 1 through 6 for no schooling, up to grade 5, up to grade 10, ��rst government examination passed, university entrance passed, and degree or diploma, respectively. Six entrepreneurship characteristics were qualitatively assessed to rank the entrepreneurship (X6) of the mill manager. These characteristics include risk perception, employee welfare technology adoption, plura-activity or diversi��cation, development of contacts and networks, and sustainability/environmental awareness. Managers/mill owners were asked a few questions on each of these aspects. Based on the answers, they were assigned a rank from 1 through 5, 1 representing very poor entrepreneurship and 5 representing very good entrepreneurship. Capacity of the mill (X7) was measured as potential to produce sawn wood per month if the mill operated for eight hours at full capacity for 22 days. If the source of energy (X8) in a mill is only electricity, that mill was assigned 1, and mills with diesel-operated machines or combination of electricity and diesel were assigned 0. In early studies of technical ef��ciency, a two-stage procedure was followed in analyzing the determinants of technical ef��ciency. In the ��rst stage, technical ef��ciency scores were estimated, and then a second-stage regression was estimated to ��nd the determinants of technical ef��ciency. As shown by �umbhakar et al. (1991), this procedure has two problems. first, technical ef��ciency may be correlated with inputs causing inconsistent estimates of the parameters and technical ef��ciency scores. Second, the standard ordinary least squares estimators are inappropriate because the technical ef��ciency scores—the dependent variable in the second stage regression—are onesided. �umbhakar et al (1991) suggest a one-step formulation in order to overcome these problems. The present study used this one-step procedure to obtain technical ef��ciency scores and their determinants simultaneously using the frontier econometric software (�ersion 4.1) program. V. REsUlts Appendix Table A2 shows the descriptive statistics of the variables used in the estimation of the frontier production function. Only the log input is signi��cant in the production function. Energy inputs and both labor inputs are insigni��cant (see Appendix Table 3). The results reveal that the log input is the limiting factor of production in sawmilling. Most of the mills are operating under capacity due to a severe shortage of logs. The poor results of the production function analysis may be due to measurement errors. had the input–output sheets been ��lled out soon after milling was undertaken, as originally planned, better results would have been obtained. The distribution of the technical ef��ciency scores is given in Table 2. The average technical ef��ciency is 0.7219. from an input perspective, this indicates that on average the sawmills can save about 28% of all inputs while producing the same output if the production process is reorganized in an appropriate manner. from an output perspective, the results suggest that on average 28% more sawn wood can be produced with the current level of inputs. Thus, the overall results indicate that there is considerable inef��ciency in the sawmilling industry. sectiOn v results erd WOrking PaPer series nO. 105 11 tablE 2 distribution of tEchnical EfficiEncy scorEs EfficiEncy scorE (pErcEnt) numbEr of mills pErcEntagE of mills < 10 1 0.68 11 – 20 1 0.68 21 – 30 8 5.41 31 – 40 10 6.76 41 – 50 12 8.11 51 – 60 3 2.03 61 – 70 6 4.05 71 – 80 23 15.54 81 – 90 41 27.70 91 – 100 43 29.05 Average ef��ciency score 0.7219 Standard deviation 0.229 Table 3 shows the factors influencing technical ef��ciency in the sawmilling industry. As mentioned earlier, the equation to analyze the determinants of technical ef��ciency was estimated using the maximum likelihood method, which does not estimate coef��cient of determination (R2). Nevertheless, the correlation between actual and predicted values is 0.68, and seven of the nine variables used are statistically signi��cant. The age of the manager/owner and the source of energy do not show a statistically signi��cant impact on technical ef��ciency. Since there are frequent power cuts, it was hypothesized that mills that completely depend on public power supply are technically inef��cient. however, results indicate that there is no statistically signi��cant relationship between source of energy and technical ef��ciency. tablE 3 factors affEcting tEchnical EfficiEncy variablE coEfficiEnt standard Error tratio Intercept –0.40366 1.58370 –0.2549 Age (X1) 0.03787 0.05036 0.7518 Quality of logs (X2) 0.35137 0.03369 10.4302** Type of mill (X3) –0.47824 0.06415 –7.4540** Owner management (X4) 0.20986 0.02942 7.1326** Education (X5) –0.22210 0.02816 –7.8855** Entrepreneurship (X6) 0.10627 0.02891 3.6756** Capacity (X7) –0.08626 0.01516 –5.6894** Source of energy (X8) 0.16516 0.11648 1.4179 Age of machine (X9) –0.28842 0.07388 –3.9040** ** Signi��cant at 0.05 level. 12 OctOber 2007 efficient technOlOgy and the cOnservatiOn Of natural fOrests: evidence frOm sri lanka herath gunatilake Quality of the log input shows a statistically signi��cant positive impact on technical ef��ciency, as expected. The mills that are only milling and charging based on the log input are inef��cient because there is no incentive for such mills to improve technical ef��ciency. As discussed earlier, mill managers are not provided any incentives based on the performance of the mill. Therefore, it is expected that owner-managed mills are technically more ef��cient. The expected relationship was observed with statistical signi��cance. Education of the owner/manager was expected to positively affect technical ef��ciency. This relationship was not observed, which may be due to lack of focus of formal education in business management. Entrepreneurship was expected to positively affect technical ef��ciency. The results show the expected impact with statistical signi��cance. As expected, there is a negative relationship between capacity of the mill and technical ef��ciency. As indicated earlier, most of the machinery in the sawmilling industry are old, and as machines get older their performance becomes poor, leading to technical inef��ciency. As hypothesized, the age of the machine negatively influences technical ef��ciency. A. Efficiency Improvements and Forest Conservation Technical ef��ciency measures the maximum rate at which the use of all inputs can be reduced without reducing outputs. It can also be measured as the rate at which the outputs can be increased with the same level of inputs (�umbhakar 1996 and Seiford 1996). As the results show, there is signi��cant inef��ciency in the sawmilling industry in Sri lanka. A data envelopment analysis was also carried out with the same data and it was found that ef��ciency scores were similar to those described above (�unatilake and �unaratne 2002). The results thus suggest that the current sawn wood output can be obtained with about a 28% cut in all inputs. Such an improvement will relieve pressure on the overall wood supply sector. however, protected forests will receive the greatest impact from the log inputs savings. As basic forest economics suggests, there is a cost gradient for illegal harvesting from different types of forests (Clarke et al. 1993). The least-cost types will always be harvested ��rst and the highest-cost types will be harvested last. Private lands such as home gardens are the least-cost wood sources while protected forests are the highest-cost sources.5 �nprotected natural forests may be in between these two categories. If current trends continue, scarcity will raise prices, providing the incentive for illegal timber extraction even from highestcost protected forests. In order to highlight the magnitude of forest savings due to ef��ciency improvements, saved logs can be converted to an area of natural forests with some plausible assumptions, as follows. The data used in this calculation were taken from MfE (1995). (i) On average the different types of mature natural forests provide the following wood volumes�� lowland rain forests – 126 m3/ha Dry monsoon forests – 21 m3/ha Moist monsoon forests – 39 m3/ha Of these forest types, lowland rainforests are con��ned to a few patches. Wood harvest from this type is not possible because they are protected. Moist monsoon forests are 5 In general, protected natural forests are the least disturbed forests and historically remained intact due to high cost of timber extraction. In addition to the location and related high cost of harvesting protected forests, there are additional costs of being caught and punished for illegal logging. Such costs are highest for the protected forests. sectiOn v results erd WOrking PaPer series nO. 105 13 also limited to small areas. Only dry monsoon forests remain in large areas. Therefore, the saved forests are calculated, assuming a weighted average of 25.5 m3 of wood can be harvested from one hectare of forests. In calculating the weights, the dry monsoon forests were assigned a weight of 3 and the moist monsoon forests were assigned 1 based on the available forest areas. (ii) As technical ef��ciency improves, the timber supply curve shifts to the right. Assuming that the demand curve does not shift, the shift in the supply curve results in a lower price and a higher equilibrium quantity of sawn wood. Thus, the technical ef��ciency improvement allows production of the extra quantity without extra log inputs. This has two effects. first, the price decrease improves social welfare because lower prices increase consumer surplus. lower prices also reduce the incentives for illegal logging. Second, producing the extra sawn wood output with the same inputs leads to saving the source of logs—and consequently the forest lands. (iii) Total round wood consumption in Sri lanka is 1,396,000m3 in 2000. Of this volume of logs, it was assumed that 28% could be saved annually if technical inef��ciency is completely eliminated. however, complete elimination of technical inef��ciency is an unrealistic assumption. Therefore, the avoided deforestation was calculated assuming that technical ef��ciency improvement are 25%, 50% and 75% of the existing inef��ciency of 28%. tablE 5 EfficiEncy improvEmEnts and prEvEntEd dEforEstation EfficiEncy improvEmEnt (pErcEnt) prEvEntEd annual dEforEstation (hEctarEs) 25 3,695.29 50 7,390.58 75 1,1085.88 Table 5 shows the extent of prevented deforestation under different levels of technical ef��ciency improvements. The assessment of the determinants of technical ef��ciency provides some avenues for technical ef��ciency improvements. Enhancing the supply of quality logs, investing in new machinery, providing hired managers incentives that are linked to ef��ciency improvements, and improving entrepreneurship abilities can be used to improve technical ef��ciency. This study analyzed technical ef��ciency improvements only at one stage of forestry—sawmilling. The analysis shows that technical ef��ciency improvement only at the milling stage has a potential role in preventing deforestation in Sri lanka. Technical ef��ciency improvements over the entire forestry life cycle (starting from tree planting up to end uses such as construction/furniture industries) may provide much higher conservation impacts. If such technical ef��ciency improvements can be realized at the regional or global levels, there would be profound positive impacts on forest conservation. The value of the ��ndings of this study is limited by the small sample size and quality of data. Moreover, technical ef��ciency in this type of studies is de��ned taking the best mills in the sample as the benchmark. If the sample is cross-country, including countries with better milling technology, much bigger inef��ciency may be discovered. Therefore, more cross-country research on technological improvements on forestry covering the entire life cycle could provide more valuable policy directions. 14 OctOber 2007 efficient technOlOgy and the cOnservatiOn Of natural fOrests: evidence frOm sri lanka herath gunatilake b. Efficiency Improvement versus other Policies The results show that technical improvement in the sawmilling industry has a potential role to play in protecting natural forests in Sri lanka. however, to better appreciate its value, comparison with other policy tools is needed. Such an analysis should answer the following questions�� (i) how effective is the policy to achieve conservation objectives? (ii) how long will the policy take to be effective?6 (iii) Is the policy measure politically feasible? (iv) Does the policy achieve the conservation objective without compromising economic ef��ciency? �unatilake and �unaratne (2002) undertook such a comparative analysis considering timber trade liberalization, and removal of timber permit system in comparison to technical ef��ciency improvement. The following provides a summary of their ��ndings. The effect of timber market liberalization on forest protection was analyzed using a static market simulation model (see Weerahewa and �unatilake 2007 for details). lack of estimated demand elasticity for timber products in Sri lanka was a constraint in this analysis. The analysis was undertaken with plausible assumption on price elasticity. Removal of existing distortions such as tariff, goods and service tax, defense levy, and other border charges can cut down the local supply of sawn logs signi��cantly. The supply would be reduced due to the impact of sawn wood prices�� removal of all border charges reduce timber price by about 25%. The decline in timber price reduces the incentives for illegal logging and enhances forest conservation in Sri lanka. The price reduction also results in an increase in consumer surplus of about �S$40 million per year. Timber trade liberalization reduces local supply by about 12% to 31% depending on the elasticity of supply and demand. Timber market liberalization is a short-run measure that can effectively reduce sawn wood prices and consequently decrease the incentives for illegal logging. Also, it will reduce the local sawn log supply and effectively lessen the pressure on natural forests. Similar to the analysis on the elimination of technical inef��ciency, the annual savings of equivalent natural forest were estimated using plausible assumptions. With the more conservative assumption of inelastic demand, timber market liberalization can save about 6,985 hectares of forest annually. If the demand is elastic, these savings can increase up to 17,469 hectares per annum. Compared to the technical ef��ciency improvement, timber market liberalization seems to have a quicker effect on natural forests. As shown in the analysis, timber market liberalization improves overall social welfare, and thus economic ef��ciency. �enerally, open market policies bene��t certain groups in the society and adversely affect certain other groups. In this case, the consumers are gainers and the producers are losers. One unique characteristic of the current forestry sector in Sri lanka is that there is no organized timber supply sector. Part of the timber is supplied from home gardens and other nonforest lands as a by-product. The rest is illegally extracted from natural forests. Timber traders and government of��cers, in both cases, appropriate most of the rents. Therefore, the actual producer surplus losses can be very low and there are no true losers in the case of timber market liberalization. however, political feasibility of timber trade liberalization is moderate. More importantly, depending on the forest management practices of the timber exporters, this policy may lead to destruction of forests elsewhere. In particular, if the timber is exported from an unsustainable source, there would be no conservation impact at the global or regional level. This seems to be the major drawback7 of the timber trade liberalization policy as a tool for conservation. 6 This requirement is unique in forestry because forestry cycles are long and the results of many policy changes are realizedThis requirement is unique in forestry because forestry cycles are long and the results of many policy changes are realized after 20–40 years depending on the forest species. for example, the effect of any incentive program for smallholder tree growers will be observed only after 20–40 years. 7 Timber certi��cation system can be implemented to prevent deforestation in the exporting country. however, costs andTimber certi��cation system can be implemented to prevent deforestation in the exporting country. however, costs and references erd WOrking PaPer series nO. 105 21 Senaviratne, J., and h. M. �unatilake. 2001. “Can Regulation of Timber Trade Protect Our forests?” Department of National Planning, Colombo. �npublished manuscript. �ictor, D. �., and J. h. Ausubel. 2000. “Restoring the forests.” Foreign Affairs 79(6)��127–44. Available�� http��//phe.rockefeller.edu/restoringforests/. Weerahewa, J., and h. �unatilake. 2007. Timber Market liberalization in Sri lanka�� Implications for forest Conservation. Sri Lankan Journal of Agricultural Economics. forthcoming. Wernick, I. �., P. E. Waggoner, and J. h. Ausubel. 1997. “Searching for leverage to Conserve forests�� The Industrial Ecology of Wood Products in the �nited States.” Journal of Industrial Ecology 1(3)��125–45. yin, R., and D. h. Newman. 1996. “The Effect of Catastrophic Risk on forest Investment.” Journal of Environmental Economics and Management 36��186–97. Zellner, A., J. �menta, and J. Dreze. 1966. “Speci��cation and Estimation of Production functions Models.” Econometrica 34��784–95. Zhang, D, J. liu, J. �ranskog, and J. �an. 1998. “China�� Changing Wood Products Markets.” Food Products Journal 49(6)��14–20. Economics and REsEaRch dEpaRtmEnt Printed in the Philippines Efficient technology and the conservation of natural Forests: Evidence from sri Lanka Herath Gunatilake October 2007 about the paper Herath Gunatilake examines the feasibility of technical efficiency improvement as an approach for forest conservation using a case study of saw milling in Sri Lanka. The paper shows that reduction of existing inefficiency helps prevent deforestation. Having compared the merits of this approach with other policy options, the paper asserts that technological improvements as a means of forest conservation deserve the attention of policymakers. Asian Development Bank 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org/economics ISSN: 1655-5252 Publication Stock No. about the asian development Bank ADB aims to improve the welfare of the people in the Asia and Pacific region, particularly the nearly 1.9 billion who live on less than $2 a day. Despite many success stories, the region remains home to two thirds of the world’s poor. ADB is a multilateral development finance institution owned by 67 members, 48 from the region and 19 from other parts of the globe. ADB’s vision is a region free of poverty. Its mission is to help its developing member countries reduce poverty and improve their quality of life. ADB’s main instruments for helping its developing member countries are policy dialogue, loans, equity investments, guarantees, grants, and technical assistance. ADB’s annual lending volume is typically about $6 billion, with technical assistance usually totaling about $180 million a year. ADB’s headquarters is in Manila. It has 26 offices around the world and more than 2,000 employees from over 50 countries. ERD WoRking PaPER SERiES no. 105