The economics of farm-level adoption of soil conservation measures in the uplands of Java

THE WORLD BANK POLICY PLANNING AND RESEARCH STAFF 6AV -cl/ Environment Department The Economics of Farn-Level Adoption of Soil Conservation Measures in the Uplands of Java Edward B. Barbier October 1988 Environment Department Working Paper No. 11 This paper has been prepared for Internal use. The views and interpretations herein are those of the author(s) and should not be attributed to the World Bank, to its affiliated organizations or to any individual acting on their behalf. The paper has Ieen prepared by Edward Barbier, a consultant to the Environment Department. The author is indebted to Gloria Davis, John English, Ernst Lutz, William Magrath, Kenneth McConnell, David Pearce, Douglas Southgate and E. Zimmer-Vorhaus for comments, as well as to participants in the World Bank study of upper watershed degradation on Java. Departmental Working Papers are not formal publications of the World Bank. They present rough and tentative results of country and sector analysis or research, and are circulated to encourage discussion and comment. Citation and use of such a paper should, therefore, take account of its provisional character. The findings, interpretations, and conolusions expressed in this paper are entirely those of the author and should not .be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Because of the informal nature of this paper and in order to present the results of research with the least possible delay, the typescript has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsiblity for the errors. ii ABSTRACT The incentives for farmers to adopt soil conservation packages as a means to controlling erosion and improving long-term land productivity has become an important topic for economic analysis. Such analysis is also relevant to the design of investment strategies for soil and water conservation, particularly to protect upper watersheds. On the island of Java in Indonesia, there has been extensive investment over the years in upper watershed management projects. Numerous technical packages for upland soil conservation have been introduced with varying degrees of success. The lessons learned from this experience has been invaluable for understanding the economics of farm-level adoption of soil conservation measures, not just in the uplands of Java but in other developing regions with similar conditions. The incentives for an upland farmer on Java to invest in soil conservation packages can be captured in a simple model relating profitability to erodibility on privately owned land. The model is used to examine how this relationship is in turn influenced by chLnges in relative prices, the farmer's discount rate, different soil qualities, slopes and depths, input prices and population, including migration for off-farm employment. Although the result of the analysis is relatively straightforward - upland farmers will not modify their land management practices and farming systems unless it is in their direct economic interest to do so - the incentives for farming households to adopt conservation packages vary significantly. For example, in some cases, such as growing horticultural crops on deep volcanic soils in the steep uplands, there is little incentive for farmers to invest in conservation as soil erosion appears to have negligible effects on farm profitability. Other incentive effects can be traced directly to government policies, such as maintaining high relative prices for cassava and vegetables, high subsidies for fertilizers and the general poor availability of rural credit at affordable rates. More complex incentive effects arise from the relationship among the availability of off-farm employment, population pressure and land management in the uplands. There is a need for more information and analysis of these different incentives. There is also insufficient data and analysis of land tenureship in the Javan uplands and its impact on land management. Exploring these and other relationships through more empirical analysis of available data is the next step in understanding the incentives for the control of soil erosion and improved land management at the farm level. iii THE ECONOMICS OF FARM-LEVEL ADOPTION OF SOIL CONSERVATION MEASURES IN THE UPLANDS OF JAVA Table of Contents Page 1. Introduction 1 2. Technical Packages for Upland Soil Conservation on Java 1 3, The Model 6 4. Comparative Static Analysis 11 5. Empirical Work and Data Requirements 17 6. Conclusion 19 References 20 Notes 23 Appendix 25 iv THE ECONOMICS OF FARM-LEVEL ADOPTION OF SOIL CONSERVATION MEASURES IN THE UPLANDS OF JAVA 1. Introduction On the densely populated island of Java in Indonesia, the area of severely eroded upland is increasing at the rate of 1-2 percent per annum and now totals around 2 million ha, approximately one third of Java's culti'vated uplands. Average population density in these areas is 600-700 people per sq km, although it may fall to 400-550 per sq km in severely degraded upper watersheds. With holdings averaging 0,4 ha or less, with up to 20-25 percent of the population being landless in some areas, and with yields for upland rice and corn averaging 0.9-2.5 tons/ha, the general pattern is one of poor, predominantly subsistence households seeking to increase their immediate basic food requirements by using inappropriate cropping patterns that result in high soil erosion levels from their rainfed lands. In addition, significant erosion problems are caused by absentee and better- off farm owners cultivating highly profitable but erosive crops, such as vegetables, and by the failure to 'police' state-owned tree plantations properly, particularly in preventing illegal fuelwood collection and agricultural conversion (Barbier, 1987; Roche, 1987; World Bank, 1987b and 1988). This paper is concerned with the incentives for upland farmers on Java to adopt soil conservation packages as a means to control erosion and improve long-term land productivity and agricultural growth. The paper begins by reviewing the soil conservation 'packages' currently offered to upland farmers and the factors influencing their adoption. An economic model is then constructed to characterize the choices open to farmers between these packages and existing cropping patterns and inputs that contribute to soil erosion. Comparative static analysis of the model can indicate the incentives for farmers to adopt soil conservation measures, as observed by existing farm-level studies. This in turn yields insights into the design of appropriate policies and incentives to facilitate the control of soil erosion on upland farms. 2. Technical Packages for Upland Soil Conservation on Java The predominant technical approach for combatting upper watershed degradation on privately owned land on Java has evolved from the FAO-funded Solo Watershed project (1972-78). In upper watershed areas up to 50% slopes, this basic approach involves subsidizing bench terrace construction through cash wages and/or free agricultural inputs, either directly or through the provision of credit. After terraces are built, food crops are planted on the 1 horizontal surface of the terrace and grasses on the riser and lip to support livestock intensification and to help control erosion. Farmers are also encouraged, through extension support, to adopt improved cropping patterns, which together with the introduction of new varieties, increased inputs (e.g., fertilizer and pesticides) and better technical information are intended to improve the net returns from cultivation. For example, in the Citanduy II project in West Java, farmers were advised not to monocrop corn and cassava, and to switch to growing higher valued crops. As a result, increased returns from terracing have been largely associated with changing crop patterns: cassava production fell from 42.4 to 12.4 percent of the average value per plot, while rice production increased from 7.1 to 26.8 percent and peanut production rose from 3.4 to 17.8 percent (Saragih, Huszar and Cochrane, 1986). However, dependence on a single technical package to reduce soil erosion and improve the livelihoods of upland farmers throughout all the watersheds of Java is unrealistic; instead, there is a growing recognition that approaches-have to be more varied, given the great variability in topographic, soil, agroecological and even socioeconomic conditions across these watersheds. For example, a study of bench terracing and related farming practices in the Citanduy basin revealed that model farms based on recent and plioone volcanic soils have the lowest soil relative erodibility index and highest net returns, whereas the model farms located on the sedimentary derived (limestone) soils have low, and in some cases negative net returns. This is due not only to the generally lower fertility of the limestone soils but also to the higher labour inputs required for implementation and maintenance of terracing and the higher rate of topsoil loss due to greater soil erodibility (Kucera et al., 1986). In addition, bench terraces may have a high risk of collapsing after a long period where there is high limestone content of the subsoil and the betonite clay mineral causes instability. Similarly, during periods of peak rainfall inappropriately bench terraced fields can lose just as much if not more soil than adjacent fields cultivated using traditional methods (nearly similar to ridge terracing) (Fagi and Mackie, 1987). The livestock component of upper watershed projects is also intended to increase household income and nutrition, reduce vulnerability to crop failure and encourage farmers to maintain a grass cover on terraces. The general approach has been for a project to provide small ruminants (sheep or goats) to participants, who are expected to return one or two of the first offspring for further distribution by the project. In the past, several projects have had difficulty in implementing a livestock credit program and/or have been unable to make animal stock available at sites; consequently, farmers have had no incentive to maintain a forage cover crop (Bernsten and Sinaga, 1983). The Kalii Konto Project in East Java, however, has had more success 2 in implementing goat and sheep schemes, which are aimed at increasing the income of the poor, landless farmers. Participants are expected to plant fuel and fodder trees in their homeyards and/or on forestland, which they could also use for their own needs. Some farmers prefer goats to sheep, because of the manure used in onion production, and because of the quicker reproduction rate and higher sales prices of goats; others have chosen sheep because they require less maintenance, can be fed with low quality fodder and are considered to be more disease resistant (Dwiwarsito,and de Graaff, 1987 p.40). On slopes greater than 50%, the basic approach has been to discourage farmers from growing annual crops and instead plant tree crops for cash income, such as clove, fruits, coffee, cocoa and tea, fuelwood for domestic consumption and sale, fodder trees and grass under the canopy for livestock. As the main constraint to developing agroforestry based systems is the long 'waiting' period before trees mature, the usual strategy, as employed in Citanduy II, has been to intercrop food crops and tree species. A preferred approach would be to develop farming systems incorporating a mix of trees that can be harvested or produce fruit at different times to spread income flows throughout the year and which gradually induce the phasing out of annual cropping over the medium and long run (Bernsten and Sinaga, 1983). In the Kali Konto project, a key component in the establishment of perennial crop gardens and agroforestry systems on steep slopes was seen to be the development of village-level nurseries backed up by central forestry and perennial crop nurseries at the (sub)district level to supply planting material required for plantings on villageland; to train and encourage farmers in the establishment of nurzeries for their own requirements; and to serve as a demonstration site for high yielding varieties. To increase the production of fuelwood, fruits and fodder in watershed areas, the provision by the nurseries of high quality seedlings in adequate quantities, at reasonable prices and at close distance to the farming population was considered to be essential (Dwiwarsito and de Graaff, 1987 ch.4). In spite of some input subsidies, the introduction of many soil conservation techniques onto upland farms may require a substantial investment of time and money by farmers. For slopes of 50% or less, to introduce bench terracing often requires a significant input of human labor, ranging from about 750 to over 1800 mandays (MD) per hectare depending on the slope. This implies construction costs of between US$ 420 and US$ 2,060 per hectare (1979 prices). In addition, costs of planting material, tools and fertilizer required to build and establish a crop on the terrace in the first year average US$ 112/ha (1979 prices). Thus the total labor and material costs would range from US$ 560 to US$ 2,075 per hectare (1979 prices). These estimates do not include the additional costs to the farmer of periodic 3 maintenance of terraces, waterways and drop structures. For farmers to add an intensive livestock system to terracing, the cost of establishing a grass cover on terraces is approximately US$ 72/ha for material (1979 prices) and an extra 2-5 MD/ha, if 20% of the hectare is in terrace risers and lips. A mature female sheep or goat costs about US$ 70 (Bernsten and Sinaga, 1983). It is generally assumed that the labor for constructing the terraces is provided.by the farmer during the dry season. Assuming a four month dry season and a holding of 0.5 hectares, a single farmer could provide a maximum of 100MD each dry season; however, this is far short of the terracing requirements of approximately 375 (low slope) to over 900 (steep slopes) MD/0.5 ha holding. Alternatively, during a 100 working day period, a farmer could only terrace 0.14 ha of low slope and 0.06 ha of steep slope (Bernsten and Sinaga, 1983). The total labor requirements for terracing may therefore mean additional cash expenditures on hired labor. Thus from the farmer's perspective, the costs of terrracing not only imply forgoing his own income- earning opportunities - either in off-farm employment during the dry season and/or less labor time devoted to crop and farm production - but also additional expenditures on material and possibly livestock costs. This suggests that adoption of terracing-based technology for soil conservation may be limited to households with available cash to hire additional labor, those with more than one adult male member to supply labor, and/or wealthier households that can afford to forgo seeking wage employment during the dry season. For example, in Gubugklakah, East Java, only the relatively wealthy farmers engaged in profitable commercial apple production (approximately US$ 3-6,000 per year profits - 1986 prices) are able to afford to construct extremely effective back sloping tied ridges capable of conserving topsoil and reducing erosions rates to less than 10 tonnes/ha (volcanic middle soils with a slope profile of 4 to 8 degrees) (Carson, 1987). Similarly, a survey of farmers who did not adopt bench-terracing technology in the Citanduy watershed of West Java revealed that 87% cited lack of money as the reason for not constructing terraces (Tampubolon and Saragih, 1986). One advantage of a livestock-based system is that owning small ruminants provides an opportunity for the household to self- employ family labor, often child labor with a low opportunity cost. On the other hand, to feed small ruminants by cutting grass along roadways and on other public land, the household must spend 1 hour/animal/day in the wet season and 2 hours/animal/day in the dry season. Under these conditions, flock size seldom exceeds 4-8 animals per household. By establishing grass intensively on terraces, households are able either to raise a greater number of animals with the same labor input or raise the 4 same number of animals with perhaps only 20% of the labor input required under the extensive out and carry system (Bernaten and Sinaga, 1983). Thus with terraces already established, a farming household can significantly raise productivity from animal husbandry with a relatively small additional investment in terms of labor and material costs. As noted above, on slopes greater than 45-50%, the recommended soil conservation strategy is for farmers to take land out of annual food crop production and adopt an agroforestry-based system to produce tre crops for cash income (e.g., clove and citrus), fuelwood for home consumption and sale (e.g., Albizzia falcata), fodder trees to supply animal feed (e.g., leucaena) and grass for livestock under canopy. For upland farmers, however, there is an additional "waiting" cost of 3 or more years to be be bourne between the initial year of land preparation and planting and the eventual maturing and harvesting of tree species. For example, although Albizia faloata is considered a relatively fast-growing tree with economic potential for fuelwood, sawn lumber and supplementary dry-season forage, harvesting cannot begin before three years for fuelwood (five years for sawn logs), and it takes five years or more for cumulative returns t exceed the initial preparation and planting costs of az-ound US$100/ha (1982 prices). Similarly, preparation and planting of Glyricidia requires 7.5 MD/ha and subsequently 3.5 MD/ha for maintenance, as well as a material cost of around US$ 45/ha (1982 prices). Farmers, however, have to wait until the third year before making the first cut, and it is not until the fifth year that harvests reach their full economic potential (around US$300/ha/yr - 1982 prices) (Sumitro, 1983). This suggests that adopting agroforestry systems may be extremely difficult for poorer farmers who are dependent on very small landholdings for food production and who have no alternative cropland or employment opportunities. On the other hand, farmers who are relatively well-off, who have sufficient lower sloped cropland to grow food and/or who have access to off-farm employment opportunities may be able to afford the "waiting cost" associated with agroforestry investments. Security of land tenure is an additional determinant as to whether upland farmers are willing to bear this "waiting cost". Hence, given that upland farmers face significant costs in adopting soil conservation measures and changes in farming systems, they are unlikely to make.changes in their land management unless there is a perceived economic advantage to doing so. In addition, the more productive or profitable the land use, the more farmers will be willing to maintain and invest in better land management and erosion control practices. Higher productivity and returns will also mean that farmers can afford to maintain terraces and other conservation structures and to continue with labor-intensive erosion control measures. On the 5 other hand, poorer upland farmers dependent on low-return cropping systems, such as maize or cassava, may be aware that soil erosion is reducing productivity but may not be able to afford to adopt conservation measures. At the other extreme, farmers with very profitable crops that are extremely erosive, such as temperate vegetables on steep upper volcanic slopes, may not consider soil conservation measures if their returns do not appear to be affected by soil erosion losses. Thus the relationship between the erodibility and profitability of different farming systems on different soils and slopes is an important determinant of whether upland farmers adcnt a soil conservation strategy. As shown in Figure 1, this relationship varies widely across Java. 3. The Model The incentives for an upland farmer on Java to invest in soil conservation packages can be captured in a simple model relating profitability to erodibility on privately owned land. It is assumed that the farmer works his land to maximize the present value of the profits stream. Thus his behavior towards soil conservation is determined by the impact of soil erosion on profits. The model can be used to examine how this impact is in turn influenced by changes in relative prices, the farmer's discount rate, different soil qualities, slopes and depths, 4nput prices and population - including migration for off-farm employment. To concentrate on these effects, however, the model necessarily ignores variations in rainfall and in other biophysical and climatic conditions. It also assumes that soil quality is constant in use; i.e., continued cultivation results in soil loss but does not reduce the fertility of the remaining .topsoil, which is assumed to be maintained by the application of additional inputs (e.g., fertilizer). Most importantly, by concentrating on the farmer's decision to invest in the control of soil loss and land degradation on privately owned and operated land, the model ignores the additional erosion problems in the Javan uplands caused by the encroachment and conversion of 'open access' and publically owned forest lands for fuelwood, fodder and shifting cultivation. The latter behavior is influenced by similar considerations as those that determine a farmer's decision in a frontier region to invest in erosion control measures or in longer 'fallows' as opposed to deforestation for cultivation (Lopez and Nicklitscheck, 1988; Southgate and Pearce, 1987 and 1988). As noted by Donner (1987, pp. 67-8), shifting cultivation is no longer of much importance on Java because of the high population density, and instead sedentary dryland cultivation prevails on upland hills and high plateaux. The model also ignores the off-site impacts of upland soil 6 erosion, such as the downstream flood damages, loss of hydropower and crop productivity, disruptions to water supply, irrigation and water transport and navigation, that result from off-site siltation, water flow irregularities and agrochemical run-off. Several studies indicate that these off-site impacts might be important in tropical and subtropical watersheds, and should be included in an economic analysis of erosion control (Easter, Dixon and Hufachmidt, 1986; Gregerson et al., 1986; Velos et al., 1985). A recent study for Java estimated the on-site crop productivity losses from man-made soil erosion on upland farms to be around US$320 million annually, compared to the additional off-site sedimentation costs of US$26 - 90 million (Magrath and Arens, 1987). The authors stress, however, the difficultly in estimating accurately these off-site impacts, particularly in separating out sedimentation arising from 'natural' geological and non-farm erosion. This paper ignores off-site sedimentation impacts not because these are insignificant but in order to concentrate on the upland farmer's decisions concerning soil conservation in the face of intertemporal depletion. Thus this paper presents a simple variant of the model developed by McConnell (1983) to reflect the choices open to upland farmers on Java as to whether or not to adopt soil conservation 'packages'. It is assumed that the upland farmer has some exclusive private rights - if not full ownership - to the land. For simplicity, also assume that only one crop is produced, or if there are multiple crops, their combined production cn be explained by a single-crop production function. This function is (1) q = f(zis, x), ti > 0, fil ( 0, f2 f0, fa2 < 0, fit f2i > 0, where q(t) is output; x(t) is soil depth; and z1(t) is a vector representing a conventional input 'package' that includes productive inputs (including labor), crop varieties and cropping patterns and techniques.' An increase in ai leads to increased output, albeit at a declining rate. It is assumed that the depth of the soil, x, also has a beneficial effect on crop production; i.e., a deeper topsoil has more nutrients and moisture to be absorbed. Eventually, though, additional soil depth adds nothing to current productivity. The positive cross partial derivatives of f indicate that additional soil depth improves the crop productivity of the traditional input package, and an increase in these inputs adds to the beneficial effects of x on crop production. However, the use of zi increases soil loss, and at an 7 accelerating rate. Thus the household has the choice of adopting an alternative input package, sa, of appropriate soil conservation methods - such as the type described in section 2 - that will conserve the soil. Note that this package often !nvolves a change in cropping patterns and even choice of crops. Itf h is defined as the change in soil depth, then (2) x 2 h(zs, za), hj S 0, hia < 0, ha > 0, has < 0, ha = ha < 0. Let p(t) be the price of the crop output, f(za, x), normalized with respect to the price of any new crops associated with s2, and ci(t) and ca(t) are the costs of the respective input packages.2 The behavior of the farming household in responee to soil erosion is therefore determined by the impact of the soil on profits. Assuming that the household works its land to maximize the present value of the profits stream, over T years this is T40o (3) PV j e-r"lpf(z,s x) - cazz - cass] dt, where r is the farmer's discount rate.a Thus it is assumed that the upland farming household will maximize (3) through choice of a and za, subject to (2) and the initial depth of the soil (4) x(O) = Xe. Given the continuous function u(t), the Hamiltonian of the problem is (5) * H = e-rt([pf(ai, x) - caZ, - c2zs] + uh(zi, a)), which yields the following first-order conditions (6) dH/dz pfi - c + uhta 0, (7) dH/dZs -C + uha = 0, (8) u ru - dH/dx = ru -pf, (9) x h(zt, za), (10) x(0) = xe, and 8 (11) lim u(T)x(T) = 0. T4 Condition (6) indicates that, for conventional productive inputs, the value of the marginal product, pfi, must be equal to the total costs. This includes not only the money costs of these inputs, o, but also their costs in terms of worsening soil erosion, uhs. Similarly, condition (7) shows that the money costs of the soil conservation package must be equated with the additional value it generates by controlling soil erosion. The costate variable, u(t), can be interpreted as the true value, or shadow price, of soil. The implicit cost of soil loss, u, must therefore grow at the rate of discount less the soil's contribution to current profits,* pfa. From (6) and (7) (12) u Z 92 Z (pfh - s) ha -hi That is, the implicit cost of soil loss must be equal to both the marginal cost of preventing soil erosion through the conservation package and the net contribution of profits of the soil degradation resulting from the use of traditional inputs. Following Lopez and Nicklitschek (1988), it is possible to characterize a solution of the problem in terms of a temporary equilibrium. As shown in the appendix, the comparative static analysis of this equilibrium depicts the two controls as functions of u, x and the parameters of the problem. The resulting functional relationships can be summarized as -4+- + +. (13) as = a1(ut p s, osl ctx) I + - + -- (14) Za = ss(u, p, a, cas,9 x) Thus, in the short run, an increase in the implict cost of soil erosion and the costs of the traditional input package will favor the adoption of the soil conservation package over st. In contrast, an increase in the relative price of the traditional crop, in the costs of the soil conservation package or in the depth of the soil, will discourage the adoption of z and favor the use of the traditional input package. From (8) and (9) the long run or steady-state equilibrium of the problem can be characterized by 9 (15) A(u,x) = u = 0 if ru pfa(z(u, x), x), > > (16) B(u,x) = x = 0 if h(aj(u, x), za(u, x)) = 0 , where in the steady state u = x : 0. The slopes of the stationary loci are given by (17) dI - hIs11 + h2sas > 0, dx IB=O hazi. + has, (18) du Pfas + p1azs < 0, dx IA= r - ptiza if fas > fasZas. One such stable solution is depicted in Figure 2, with (us, x*) representing the steady-state equilibrium of the system. Thus, the optimal policy of the farming household depends on whether the initial topsoil level is high (xo > x*) or low (xo < x*). If initial soil depth is extremely high - which seems to be the case in those areas of the Javan uplands with extremely deep volcanic soils - the optimal strategy for the farmer is to deplete the topsoil until the costs of soil erosion rise. It is only when the productivity impacts of soil erosion are significant that farmers will invest in a soil conservation package to halt erosion. In contrast, if initial soil depth is poor, the value of additional crop production associated with soil conservation exceeds its costs. The problem for most upland farmers on poor soils, however, is that the low productivity of these soils may mean they do not have sufficient capital to invest in conservation packages. Nor can they afford the additional risk involved in changing their farming methods and cropping systems. Two villages in the uplands of East Java illustrate these extremes: In Srigonco, the shallow limestone soils require an agroforestry approach to soil conservation based on introducing deep rooted perennial, horticultural and agrofroestry species. Although the predominantly poor and small farmers are very much aware of the loss in productivity from soil erosion, many cannot afford the initial investment nor take on the extra risk needed to improve their existing cassava and cassava-maize farming systems. This attitude is often reinforced by lack of secure land tenure, and in some areas, a lack of interest in agriculture with increasing off-farm employment. For example, the badly managed, unterraced lands were invariably those not owned or controlled by the farmer but under a precarious lease system 10 (tanah marinir). If profitability of maize and cassava drops too low, farmers will make the rational decision to leave the farm for urban areas, where labor has better returns than on the farm. In contrast, in Ngdadas, because of deep volcanic topsoils, commercial vegetable farmers (mainly potato but also onions, garlic and cabbages) do not appear to experience significant declines in yields despite average annual erosion rates of 160- 200 tonnes/ha and the loss of 2 cm of soil each year. Soil conservation measures are therefore not perceived to be a high priority. Nevertheless, farmers rely on their land for all income and thus indicate a willingness to make land improvements if there is a demonstrable impact on productivity (Carson, 1987). 4. Comparative Static Analysis Further insights into the economic incentives for upland farmers on Java to invest in soil conservation can be gained through comparative static analysis of the model. In this section, we will analyze changes in farmers' discount rates, relative prices, input costs and population. A permanent increase in the discount rate, r, can be interpreted as an increase in the marginal opportunity cost of capital for farming households. An increase in r is therefore tantamount to a rise in the cost of credit available to the farmer for investment.4 The effect of an increase in r on the system can be seen from (15) (19) dA u > 0. dr As (18) shows that dA/du > 0, then the curve A = u = 0 must shift down in the ux plane. As depicted in Figure 3, in the new long- term equilbirium both soil depth and its shadow price are lower. Thus one impact of the rise in the farmer's discount rate is that it makes the household more myopic with regard to the future profitability of the land. This will mean a bias towards investing in traditional inputs to boost current productivity at the expense of conservation investments to halt soil erosion and improve future productivity. Eventually, however, the decline in the soil depth over time from this decision will mean lower productivity in the long run. In addition, higher lending rates clearly affect farmers' ability to invest in upland soil conservation measures. For example, farm-level investment in upland soil conservation measures particularly requires medium and long-term credit at affordable rates. Experience in the Citanduy II project indicates that returns form selected agroforestry and bench terracing farm package investments are sufficient to pay loans at 24% annual 11 interest. Investments in terracing, however, require a medium- term loan for at least 2 years and a short-term loans for succeeding years. After 9 years of investment in terracing and 12 years of investment in agroforestry, all loans could be repaid. Moreoever, the increase in net benefits after 10 years of investment supported by credit is 3 and 4 times greater respectively for terracing and agroforestry when compared to the net benefits of farms that do not adopt soil conservation measures (Tatuh, 1987). At present, neither the availability of rural credit nor its *allocation is adequate in all of Indonesia, let alone in the uplands of Java. Despite an implicit agricultural credit subsidy of about US$ 80.3 mn annually (1986/87 prices) that provides loans at 12% a year, public liquidity credit is estimated to meet only 15% of the demand for credit by farmers, and the other 85% is obtained informally at an interest rate of around 60% per annum. Indications are that small farmers in particular must rely on high cost, informal sources of funds (World Bank, 1987a). Such high interest rates are a disincentive to medium and long-term investment by upland farmers in soil conservation measures and thus encourage a bias towards short-run production decisions that leave the soil vulnerable to erosion. Comparative static analysis can also indicate the general effect on farmers' decisions of a change in the relative price, p, of traditionally cultivated commodities. Hence, from (15) and (16) (20) dA - f2 - pf2zzP < 0, dp (21) dB = hiz, + h2zp < 0, dp and as (17) and (18) show that dB/du > 0 and dA/du > 0, both the B = 0 and the A = 0 curves shift up (See Figures 4 and 5). Thus in the long run, the cost of soil erosion will rise unambigusouly as the relative price of output produced with tradtional productive inputs, a, rises. That is, the resulting demands on the productivity of the soil of the stimulus to current production and the additional soil degradation incurred over the long run increases the value of soil depth. Whether in the long run, soil depth itself will decrease, remain the same or increase will depend on whether the negative impact on soil erosion of an increase in p will exceed its positive impact on the current productivity of traditional production methods; i.e., it will depend on whether the B = 0 curve shifts up more than the A = 0 curve (22) dx* 0, if hiz, + htz,p. f2 + pfa1zp .12 Figures 4 and 5 depict respectively the cases where the B = 0 curve shifts up more and less than the A = 0 curve. It is easy to demonstrate that the effect of a decline in p, i.e. an increase in the relative price of output produced with the new cropping systems introduced with the conservation package, za, will have the converse effect, as depicted in Figures 6 and 7. Consequently, the impact of a change in relative prices is significantly influenced by the effect of this change on soil degradation as opposed to current productivity. This effect may in turn be determined by the characteristics of the soil in different upland conditions. For example, on the poor quality, shallow limestone-based soils found in certain upland areas, the marginal productivity of the soil, ft, is generally low and the boost to soil productivity of using additional productive inputs, fal, is close to zero. On the other hand, the impacts of these inputs on soil degradation, hi, are very substantial. As a result, condition (22) would suggest that the effect of an increase in p on these soils would be a drastic decline in soil depth over the long run (see Figure 4). Thus on poor quality, highly erodible soils the success of projects in encouraging farmers to adopt conservation packages is often determined by whether these packages allow farmers to shift to higher valued crops. For example, in the Citanduy project, it has been observed that, as one of the functions of terracing is to conserve soil moisture, such conservation measures have induced farmers to shift from low-value crops such as cassava to crops that earn three to five times more per kilogram. Over a three-year period, cassava production fell from 42.4 to 12.4% of the average value per plot, whereas rice production increased from 7.1 to 26.8% and peanut production rose from 3.4 to 17.8%. In addition, the mix of inputs (i.e., labor and purchased inputs such as fertilizers, seeds and pesticides) was observed to be sub-optimal before terracing, but after terracing approached the optimal mix (Saragih, Huszar and Cochrane, 1986). In recent times, however, this policy of discouraging the mono-cropping of cassava because of its deleterious impact on soil structure has been 'undermined by the quadrupling of cassava prices, which has been encouraged by the Government of Indonesia's current pricing, targetting and export-promotion policies (World Bank 1987b). In contrast, on the deep, highly productive volcanic soils, soil productivity and its response to increased productive inputs are very high. Extremely deep topsoils also mean that it takes considerable time before soil erosion severely reduced soil productivity. Consequently, as indicated by condition (22), until this occurs, the long-run impact on productive topsoil depth may be negligible. As discussed in section 2, such soils are considered ideal for growing high-valued horticultural crops, although there is increasing concern over the off-site impacts of 13 the high erosion rates associated with their cultivation. Over the period 1976-86 in Java, the farmers' terms of trade for growing vegetables and their price relative to other crops have increased substantially, due to Government import controls and considerable demand in urban areas. As a result, upland vegetable farmers have little incentive to invest in conservation measures, and in fact, the production of vegetables in upland areas has expanded considerably. On the other hand, high effective protection rates for fruit and other perennial crops encourage the switch to conservation methods based on agroforestry (Roche,'1987 and World Bank, 1987b). The effects of an increase in the costs of the respective input packages can also be analyzed. For example, the impact of an increase in the cost of the traditional input package, c, is (23) dA = - pfazie > 0, dcI (24) dB = hiszit + hiaci > 0, dcI which indictes that both the A = 0 and B = 0 shift down (See Figures 6 and 7). Over the long run, an increase in the costs of the traditional productive inputs will mean less demands on the soil from current production and reduced soil degradation. Thus the aost of soil degradation will fall. Whether soil erosion will actually be reduced or increased over the long run again depends on the impact on reducing soil degradation relative to the effect on soil productivity (25) dx* 0 if hize, + ha2sC - pfazsct. Thus on poorer quality soils where the impact of increased productive inputs on soil productivity is negligible, i.e. faj40, yet their impact on soil degradation, hi, is high, one would expect an increase in oi to be an incentive for investments to control soil erosion over the long run (See Figure 6). Conversely, there is concern that the access to certain subsidized inputs by upland farmers might prove to be a disincentive to control the long-term productivity damage of soil erosion. This is particularly the case for fertilizers, the increased application of which may artifically boost yields in the short run on even severely degraded soils. With current subsidies for fertilizers in Indonesia estimated to be about 38% of the farmgate price (World Bank, 1987a), upland farmers are encouraged to apply relatively cheap fertilizers to increase yields rather than consider more expensive but environmentally sound methods such as green manuring, mulching and using compost to maintain soil fertility and stability. For example, in 14 Ngadas, East Java, farmers are presently using over 1,000 kg of subsidized chemical fertilizers per heotare to produce two 10- tonne potato crops. These yields are less than one half of what could be attained with improved soil management techniques; as the farmers have only recently realized that the reduction in yields from soil erosion were not being offset by the massive fertilizer dosages, they are now increasing their use of organic fertilizers (Carson, 1987). Similarly, an increase in the costs of soil conservation inputs, c, shifts up both the A = 0 and B = 0 curves, as (26) 4A = - pfaZc2 < 0, dC (27) dB = haszea + h2ZaC < 0. dc2 Consequent ', as protection of the soil proves more expensive, its value over the long run will increase (See Figures 4 and 5). The relative impact of increased soil degradation compared to any current boost to soil productivity from using more productive inputs again determines whether topsoil depth declines significantly over the long run; i.e. (28) dxs = 0 if hazics + haZiC - pf2lZC In general, for most of the uplands on Java, one would expect a long-run increase in soil erosion, as depicted in Figure 4. Thus it is clear why most upland soil conservation projects have chosen to subsidize some of the inputs, particularly for farm labor, needed for conservation measures (see section 2). Such subsidies are considered important in encouraging the poorer farmers who cannot afford the conservation investments or the added risks these investments pose. For example, in Gubugklakah, East Java, only the relatively wealthy farmers engaged in profitable commercial apple production (in 1986, profits of approximatley US$ 3-6,000 per year) are able to afford to construct extremely effective backsloping tied ridges capable of conserving topsoil and reducing soil erosion rates to less than 10 tonnes/ha on the volcanic middles soils with a slope of 4 to 8 degrees (Carson, 1987). Similarly, a survey of farmers who did not adopt bench-terracing technology in the Citanduy watershed of West Java revealed that 87% cited lack of money as the reason for not constructing terraces; on the other hand, nearly two thirds of the supposed "control group" outside of project areas and a significant number of non-subsidized farmers within these areas seem to be spontaneously adopting and adapting the bench-terracing based 15 soil conservation packages. Although their use of non-labor inputs is less than project farmers who are subsidized for these inputs, these "spontaneous" adaptors have nevertheless significantly increased their income and appear to have even outperformed the more recent model and expansion farmers (Tampubolon and Saragih, 1986). Thus subsidies of conservation packages may not always be necessary if farmers can perceive an immediate economic gain from adopting these measures. Calculations of the net present value of the gains from terracing in the Citanduy II project suggest that, if terracing costs are around US$500 or lesi*(1984/85 prices), farmers could be expected to adopt terracing without subsidies (Saragih, Hussar and Cochrane, 1986). Indicating the effects of population growth and the migration to off-farm employment opportunities can be simulated though comparative static analysis by assuming that changes in productive inputs, dzi, and in conservation inputs, dza, represent changes in the input of labor in cultivation and soil conservation respectively.$ Thus the impact of population growth on the incentives for upland farming households to control soil degradation depend on the extent to which the additional labor is allocated to cultivation as opposed to soil conservation. In general, in predominantly subsistence agriculture, one would expect that increased population pressure would place a greater stress on cultivation, producing the following effect (29) dA = - pfst < 0, des (30) dB = hi < 0 aZ, As indicated in Figures 4 and 5, the increased stress on the topsoil should over the long run increase its value. With the generally poor quality upland soils, one would expect soil erosion to increase over the long run (See Figure 4), as (31) dx* 0 if hi -pfa Thus one would expect any reduction in population pressure in the uplands of Java, through increased off-farm employment activity, out-migration and reduced birth rates over the long run, to be beneficial to controlling problems of soil erosion. On the other hand, as many soil conservation methods are labor-intensive, and the purchase of labor inputs may involve a considerable cash outlay, the lack of farm labor available for conservation works may be a considerable constraint. The availability of off-farm employment may also act as a disincentive to further on-farm investments to control erosion and improve soil productivity; 16 farmers with a large proportion of their income from off-farm employment may respond to reduced profitability of their agricultural land by devoting more labor to off-farm activities. In general, the positive effect of an increase in labor available for soil conservation efforts can seen by (32) dE = hs > 0, dz2 which indicates that*the B = 0 curve shifts down (See Figure 8). Consequently, more labor available for soil conservation reduces soil degradation and its marginal costs over the long run. On the other hand, in overpopulated areas, one would expect that ha40. Moreover, although increased off-farm employment may mean upland farming households are less attached to their lands and may not invest in needed improvements, long-term trends of out- migration in search of off-farm employment may have a beneficial impact on upland areas. For example, in Belik, East Java, a significant number of the village's permanent residents migrate seasonally to nearby towns to obtain work in cigarette factories, bakeries, and as construction workers, small traders and domestic servants. For many families, this migration is quasi-permanent and only occassional return visits are made to Belik. If these demographic trends continue over the next 10-15 years, both the labor force and thus intensity of land use of Belik will decline. As a result, there may be considerable potential for new perennial-based farming systems which have lower labor requirements than current food crop systems and generate less soil erosion (KEPAS, 1985 p.95). The possibility of such trends has led some analysts to conclude that off-farm employment generated by urbanization on and off-Java in Indonesia is potentially having a beneficial impact as an alternative to more intensive cultivation of fragile lands, and as a means for the rural poor to improve their incomes and living conditions (Williams, 1987). In the case of Java's uplands, there is insufficient data and analysis of the positive and negative impacts of trends in off-farm employment on land use practices to test such a hypothesis. 6. Empirical Work and Data Requirements The relationships and parameters developed by the above model have only been partially explored by the empirical work discussed throughout this paper. For example, studies at Citanduy II have explored the economic returns to farmers of adopting the model farm conservation package and compared these to the returns received by control groups, but there was no attempt to measure either the user cost of soil degradation, u, or changes in soil depth, x (See, for example, Saragih, Hussar and Cochrane, 1986). 17 A recent re-evaluation of these studies suggests that, in any event, the 'independence' of the control group may have been compromised as it appears that many farmers in this group had been spontaneously adopting and adapting the model farm package (Cochrane and Huszar, 1987). Nevertheless, estimating the user cost of soil erosion, u, may be possible from both micro and macro-level data. At the household level, it may be estimated from analysis of the increase in farmers' incomes from.adopting soil conservation packages. For example, based on the results of the GOI's Greening and Reforestation program, a rough estimate of the user cost of upland erosion is US$155 per hectare per annum (Barbier, 1987). Alternatively, the calculation by Magrath and Arens (1987) of the on-site productivity costs of soil erosion for the uplands of Java could be interpreted as an aggregate estimate of the shadow price of soil loss. Averaged over all arable dryland on Java, it amounts to US$68/ha per year. The first step in a more comprehensive analysis would be to estimate the functional relationships indicated by equations (13) and (14) for a given project or upland region. This would indicate how changes in the user costs of soil erosion, soil depth, relative prices and relative costs influence the decisions by farming households to adopt soil conservation packages, za, over traditional input packages, a. Data on relative costs and prices should be easily obtained from local or district markets and household surveys. The user costs of soil degradationcould be estimated from surveys and analyses of sample plots, with the above aggregate estimates serving as references. Soil depth changes for sample plots would have to be continuously monitored and analyzed, or perhaps preferably, farm lands and plots for which soil loss data has already been recorded and analyzed should be included in the study area. It would be desirable to extend the analysis to include the impacts of off-farm employment. For example, one hypothesis that should be tested is that farmers with a large proportion of their income from off-farm employment may respond to reduced profitability of their agricultural land by devoting more labor to off-farm activities. Important additional data required would include information on off-farm wages, the seasonality of off- farm employment and the proportion of off-farm to total income. Details on the sex and age composition of household off-farm labor and the role of off-farm income in supplementing on-farm investments would also be useful. The analysis could also be extended further to incorporate the role of changes in marketing/transportation facilities, the availability of credit and population size and composition. In general, economic and social conditions will determine which additional relationships require inclusion in the analysis. 18 6. Conclusion This paper has explored the incentives for upland farmers on Java to invest in soil conservation packages as an alternative to existing methods of cultivation that lead to considerable soil erosion. The model illustrates, with the use of supportive evidence, that upland farmers will not modify their land management practices.and farming systems unless it is in their direct economic interest to do so. The failure of farming households to adopt conservation packages can therefore be explained by the lack of sufficient economic incentives. In some oases, such as growing horticultural crops on deep volcanic soils in the steep uplands, there is little incentive as soil erosion appears to have negligible affects on farm profitability. Other incentive effects can be traced directly to government policies, such as maintaining high relative prices for cassava and vegetables, high subsidies for fertilizers and the general poor availability of rural credit at affordable rates. More complex incentive effects arise from the relationship among the availablity of off-farm employment, population pressure and land management in the uplands. In general, there is a need for more information and analysis of these different incentives. There is also insufficient data and analysis of land tenureship in the Javan uplands and its impact on land management. The few village surveys conducted tend to confirm the general observation that farmers without security of tenure over their fields are interested in maximizing their short-term investment in seed, fertilizer and labor for the crop that is in the ground. In most cases, the choice of farming systems is determined by landowners, who are increasingly absentee from the immediate farming areas and are also primarily concerned with immediate returns from the land. Thus, in both middle and upper volcanic areas on Java, share tenancy and absentee ownership are becoming more common with the increasing profitability of fruit crops, sugarcane and vegetables (Roche, 1987). For example, in Sumberbrantas, East Java, nearly half of the land is owned by families that do not live in Sumberbrantas, which tends to be leased out to local farmers for mainly potato growing. The major consequence is that there is relatively little interest shown by farmers in land conservation, and existing terrace management and elephant grass maintenance is inadequate. This lack of interest is reinforced by the high short-term profits associated with potato growing in the village (KEPAS, 1985 pp.170 and 176). If the control of soil erosion and improved land management in the uplands of Java are to be a priority, then policies should be enacted that reinforce incentives for their adoption at the farm level. 19 High ' is. a Potatoes, sloping land o Maize-cassava o Tobacco upper volcanic nonterraced sloping, upper volcanic z z o Closed teak plantation u) 10o slope o Fruit orchard Mod. intercropped a o Maize-cassava terraced Z o Sugarcane o Fruit orchard terraced, grass covered o Grassland Low o Natural o Triple rice cropping forest Low Moderate High RELATIVE PROFITABILITY Figure 1. The relationship between relative profitability and amount of erosion of major crops, Java uplands Source: Carson and Utomo (1986), Figure 4. - 20 - ******* * -en m a snn***dP U Figure 2. The phase diagram solution Figure 3. The effect of AaO shifting down .-21-- s o Figure 4. The effect of BO shifting up more than A=O g'co Figure 5. The effect of 8=0 shif:ting up less than A=0 - 22 - I A. I Ix Figure 6. The Effect of Ba0 shifting down more than A=O A #go Figure 7. The effect of 6=0 shifting down less than A=0 - 23 - Figure 8. The effect of 8=0 shifting down - 24 - Notes 1. To simplify analysis, the production function, f, is assumed to be in 'per heetare' terms. Thus we are necessarily talking about a representative upland household on an average-size plot. Intuitively, and suggested by the evidence presented in section 2, changes in the size of the plot can affect the farmer's decision to adopt different cropping systems that conserve the soil better. For example, a farming household with a larger holding is more likely, ceteris Parabis, to take the risk of trying out tree crops on some of its land as the remaining land may still be adequate to grow sufficient food. On the other hand, even a relatively large holding may be so fragmented into smaller parcels in different locations that it is not economical to cultivate tree crops. It is possible to incorporate these considerations into the model, although it is not done so in this version. 2. Where crop outputs or inputs are non-marketed, such as food produced for subsistence, in-kind inputs, labor exchange, etc., p(t), ot(t) and oa(t) represent the relevant shadow prices and costs respectively. 3. Note that in his model, McConnell (1983) also includes a terminal resale value of the farm, R(x), as part of the net value of the farm to be incorporated in the objective function. Thus McConnell argues that a strong bequest motive or a smoothly functioning capital market induces the farmer to value R(x) as part of his income stream. The capital markets in upland Java, however, do not operate 'smoothly', as evidenced by informal lending rates of around 60% per annum (World Bank, 1987a). Similarly, with hereditary land rights and tenure arrangements, and with virtually all members of the household having a 'stake' in the current and future profitability of the land, the household may be more interested in the present value of the stream of profits in perpetuity rather than its resale value by a certain time T. This is the assumption underlying the model here. 4. However, there may be complicating factors caused by the absence of a credit market. For example, if farmers are unable to borrow, then there is no longer a relation between the rate of time preferences of farmers and their marginal value of capital, unless there is some sort of internal market for allocating funds within each farm household. Further analysis is therefore required to establish any such relationship. I am grateful to Kenneth E. McConnell for pointing out this complication. 6. Alternatively, an explicit constraint on the supply of labor could be included in the model, some of which would be allocated to cultivation as opposed to conservation. This would yield the same results at the margin, but would show more explicitly the - 25 - relative labor demands and costs of the two packages. I am, again, grateful to Kenneth B. McConnell for this suggestion, which is left for the interested reader to pursue. Such an approach could be usefully extended to include the allocation of labor to off-farm employment. The following comparative static analysis attempts to establish the important relationships between labor use and soil conservation, albeit in a more indirect fashion. -26- Appendix Comparative Statics of the Temporary Equilibrium Equations (6) and (7) in the text represent the first-order conditions of a static version of the model. Total differentiation of (6) and (7) yields fidp + pfiadx + pfaidzi - do, + hidu + uhaidzi + uh12dz: = 0. -des + hadu + uh:dz + uh*ada 0. Re-arranging terms and writing in matrix form, one obtains H dzi = dol - h1du - fadp - pfisdx dzs dox - h2du where H is the Jacobian matrix given by H pffi + uhi& uhias uha uhas The determinant of H can be signed as H I = Uh2a(pfit + uhi ) - u2hahas > 0, which is required as part of the second-order conditions for maximization of the Hamiltonian (5) with respect to the control variables, za and z. Using Cramer's rule, one can therefore obtain the effect of changes in u, p, 01, c2 and x on each of the control variables dza/du -hiuhas - (-hauhL*) 0. IHI dza/du -h2(pfa + uhia) + uhatht > 0. IHI dzi/dp = -fuhas > 0. -,27 - dzn /др = f i иhп i < 0. Н1 dzt /двi = иhп п < 0. � дгп /де1 = иhп i > 0. Н1 ds1 /деп = uh,1 п" > 0. 1Н1 • . дzп /деп = p�i 1+ uhi i t 0. ----��----- dza jdx = -pfi п иhп п >� 0. 1Ht дzп /дх - f 1 п uhs i < 0. These results аге summarized by equationв (13) апд (14) in the text. - 28 - References Barbier# Edward B. 1987. "Natural Resources Policy and Economic Framework"# Annex 1 in J. Tarrant et al.. Natural Re&ources and Environmental Management in Indonesia, USAID, Jakarta, Indonesia. Bernstenj R. and Sinaga, R. 1983. "Economics"# Technical Appendix VIt Government of Indonesia/USAID# Composite Report of the Watershed Assessment.Teamg Jakarta# Indonesia. Carson# Brian. 1987. A Comparison of Soil Conservation Strategies in Four-Agroecological Zones in the Upland of East Java, KEPAS# Malang, Indonesia, July. Carson# Brian and Utomo, Hadi. 1086. Erosion and Sedimentation Processes in Java, KEPASt Malang, Indonesia. Cochraneq Harold C. and Huszar, Paul C. 1987. Economic Analysis of the Model Farm Program and Its Subsidization under the Citanduy Il ProJects Preliminary Report, USAID and USESE, Jakartat Indonesia# August 20. Donnert Wolf. 1987. Land Use and Environment-in Indonesia. C. Hurst & Cot London. Dwiwarsitot Rurianto and de Graafft Jan. 1987. Economic Impact of Watershed Development Activities at the Village Levelp Kali Konto rojectj Malang, Indonesia, August. Eastert K. Williamt Dixont John A. and Hufachmidt, Maynard (Edo.). 1986. Watershed Resource Management: An Integrated Framework with Studies from Asia and the-Pacifict Westview Press, Boulder, Colorado. Fagi, Achmad M. and Mackie, Cynthia. 1987. "Watershed Management in the Uplands of Java: Past Experience and Future Directions"t Paper presented at "Soil and Water Conservation on Steep Lands"t Soil Conservation Society of Americat San Juant Puerto Ricog March'22-27. Gregersen, H.M.9 Brooks$ K.N.t Dixon, John A. and Hamilton, Lawrence S. 1986. Guidelines for Economic AMraisal of Waterhaed Management ProJectsq FAO# Rome, Italy, March. KEPAS. 1986. The Critical Uplands of Eastern Java: An Agroecosystem Analysis# KEPASs Agency for Agricultural Research and Developmentt Jakartat Indonesia. Kuceraq Karlt et al. 1986. Micro Model Farm Assessment of Land Resources, Directorate General of Reforestation and Land Rehabilitation, Dept. of Forestry and USAID, Citanduy, Ciamist -'29 Indonesia, May. Lopez, Ramon and Nicklitschek, Mario. 1988. "Notes for a Research Proposal on the Interactions of Natural Resources and Agricultural Growth in Sub-Saharan Africa: Trade Policy Implications", draft proposal, World Bank, Washington, DC, February. Magrath, William B. and Arens, Peter. 1987. "The Costs of Soil Erosion on Java - A Natural Resource Accounting Approach", World Resources Institute, Washington, DC, November. McConnell, Kenneth E. 1983. "An Economic Model of Soil Conservation", American Journal of Agricultural Economics, Vol. 65, No. 1., February, pp. 83-89., Roche, Frederick C. 1987. "Sustainable Farm Development in Java's Critical Lands: Is a 'Green Revolution' Really Necessary?", Division of Nutritional Sciences, Cornell University, Ithaca, new York. Saragih, Bugaran, Huszar, Paul C. and Cochrane, Harold C. 1986. "Model Farm Program Benefits: The Citanduy Watershed", USAID, Jakarta, Indonesia, July. Southgate, Douglas and Pearce, David W. 1987. "Natural Resource Degradation in Developing Countrie: A Causal Analysis of Agricultural Colonisation", Discussion Paper 87-26, Department of Economics, University College London, London. Southgate, Douglas and Pearce, David W. 1988. "Agricultural Colonisation and Environmental Degradation in Frontier Developing Countries, draft paper, Department of Economics, University College London, London, February. Sumitro, Achmad. 1983. "Tree Crop Management", Technical Appendix V, Government of Indonesia/USAID, Composite Report of the Watershed Assessment Team, Vol. 3, Jakarta, Indonesia. Tampubolon, S.M.H. and Saragih, Bungaran. 1986. "Model Farm Upland Farming Technology in the Citanduy River Basin; A State of the Art", USESE, Ciamis, Indonesia. Tatuh, Jen. 1987. "Credits for Soil Conservation and Dryland Farming Development in Upper Citanduy Watershed Region", USESE, Ciamis, Indonesia. Veloz, Alberto, Southgate, Douglas, Hitzhusen, F. and Macgregor, Robert. 1985. "The Economics of Erosion Control in a Subtropical Watershed: A Dominican Case", Land Economics, Vol. 61, No. 2, May, pp. 146-155. - 30 - Williams, David. "Indonesia: Investment Strategies, Management of Natural Resources, The Urban Sector", draft paper, World Bank, Washington, DC, April. World Bank, 1987a. Indonesia - Agricultural Incentive Policies: Issues and Options, Vol. I: The Main Report, World Bank, Washington, DC, July. World Bank, 1987b. Indonesia - Java Watersheds: Java Uplands and Watershed Management, World Bank, Washington, DC, November. World Bank, 1988. Indonesia - Forests., Land and Water: Issues in Sustainable Development, World Bank, Washington, DC, March. - 31 -