The costs of soil erosion on Java : a natural resource accounting approach

THE WORLD BANK POLICY PLANNING AND RESEARCH STAFF Environment Depaitment The Costs of Soil Erosion on Java: A Natural. Resource Accounting Approach William Magrath PeterArens August 1989 Envirment Department Working Paper No. 18 Ihis paper has been prepared for intal use. The views and intrpretatons herin are ose of the author(s) and should not be attnud to the World Bank. to its affiliated orgaizations or to any individual acting on dthir behalf.  The Authors are, respectively, Environmental Specialist in the World Bank's Environment Policy and Research Division, and Soil Science Consultant, Wageningen, The Netherlands. The report was prepared as a background report to a World Bank study of environmental concerns facing Indonesia and as part of an ongoing World Resources Institute program of research on methods of implementing the natural resource accounting concept. Support and advice from staff of the Government of Indonesia from Robert Repetto, Richard Ackermann, Dirk Leeuwrik and Gloria Davis are gratefully acknowledged. Glenn Morgan and his colleagues in the Bank's Center for Earth Resources Analysis implemented the geographic information systems model. In Indonesia, the intercessions of the State Minister for Population and Environment, Kmiil Salim, allowed us access'to important sources of data. Heri Sailo's, Kathy Harrington's and Olivia McNeal's toleration at typing repeated revisions is much appreciated. Any errors remaining in the analysis are entirely the responsibility of the authors., Departmental Working Papers are not formal publications of the World Bank. They present preliminary and unpolished results of country analysis or research that are circulated to encourage discussion and comment; citation and the use of such a paper should take account of its provisional character. The findings, interpretations, and conclusions 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 informality and 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 responsibility for errors.  - ii - ABSTRACT Soil erosion is analogous to the depreciation of man made assets. Unlike the depreciation of capital assets, however, the effects of soil erosion are not reflected in conventional measures of economic welfare. This occurs because efficient markets seldom exist for soil resources, because of the pervasive infiuence of externalitites on the true costs of soil erosion, and because systems of national accounts are biased to treat natural resource as free goods. As a result, policymakers do not have the information required to adequately weigh the benefits and costs of alternative soil conservation policies. ,he basic requirements for calculating the on-site costs of resource degradation are understanding the dimensions of the physical processes of change, understanding the impact of those processes on the production of valued goods and services, and understanding the ways in which economic activity adjusts to these changing circumstances. for this study, these requirements were met by developing three linked models. To satisfy the first requirement a geographic information systems based model was used to integrate data on soil type, topography, rainfall and landuse to estimates of levels and distribution of erosion. To estimate the productivity consequences of erosion a model was developed focussing on rainfed agricultural land. Finally, an economic model of farms response to falling productivity and of farm profitability is used to value the erosion process. The deposition of soil at downstream locations frequently reduces the benefits from investments in infrastructure such as reservoirs and irrigation systems. An effort was made to identify major categories of potential damage and to locate whatever evidence was available on their economic significance. For Java, as a whole, it is estimated that erosion costs the economy between $340 and $406 million per year. Of this $315 million are estimated to be on-farm losses of productivity and the balance $25-8 are of downstream damages.  - iii - Table of Contents THE COSTS OF SOIL EROSION ON JAVA -- A NATURAL RESOURCE ACCOUNTING APPROACH Introduction .... ...................................... 1 I. Measures of Land Degradation on Java ................ ..... 2 II. The On-site Costs Of Soil Erosion ....................... 3 A. Estimating the Physical Dimensions of Soil Erosion ....... 3 B. Estimating Productivity Effects of Erosion .............. . 8 C. Estimating the Economic Implications of Productivity Declines ................................................. 20 IlI. Off-Site Costs of Soil Erosion .......... .. ............. 28 A. Siltation of Irrigation Systems .......................... 31 B. Siltation of Harbors and Dredging ........................ 37 C. Reservoir Sedimentation .................................. 37 D. Other Off-Site Costs of Erosion .......................... 47 IV. Summary ................................................. 47 REFERENCES ............... ................................ 52 Appendices  Introduction Soil erosion is both a physical and an economic process. The physical removal of part of the topsoil and its deposition elsewhere lowers the agricultural potential of a site and thus sets in motion a sequence that ultimately results in a lower ecoromic value of the resource base. Unlike the depreciation of other capital assets, the effects of soil erosion are not normally reflected in measures of economic welfare. This occurs because efficient markets seldom exist for soil resources and because of the pervasive influence of externalities on the true costs of erosion. As a result, policymakers do not have the information required to weigh the benefits and costs of alternative soil conservation policies. In this paper a natural resource accounting approach is used to quantify in economic terms the cost to the economy of watershed deterioration as manifested in soil erosion. This analysis enables policymakers to compare the consequences of upland deterioration with other developments in the economy.1 To estimate the economic significance of soil erosion it is necessary to develop a model of the physical dimensions of erosion, link these to changes in crop production and farming systems or the production of other goods and services, and finally value these changes. In Section II the process by which the on-site costs of soil erosion on Java were estimated is described. The methodology involves use of a computerized geographic information system (GIS) in which the size of areas equally susceptible to erosion are quantified. Estimates of these levels of erosion and agronomically based estimates of the impact of these levels on crop yields are then combined with data on the predominance of alternative upland farming systems. This yields ectimates of reductions in agricultural output due to erosion. Representative farm budgets are used to value those changes. The capitalized sum of the predicted reduction in returns to land that results from this procedure provides an estimate of the on-site cost to the economy of soil erosion. Additional details on the various steps in the methodology are provided in this section along with summaries of the data generated in the process. In addition to the on-site costs of soil erosion, an attempt is made in Section III to calculate the level of major off-site or downstream costs. These include reservoir and irrigation system siltation and siltation of harbors and waterways. Less data is available on the physical dimension of the off-site consequences of erosion. However, it is possible to get an indication of their economic significance by examining data on maintenance expenditures that are necessary to ameliorate the downstream deposition of silt, or by extrapolating from particular studies of specific investment projects. 1/ For an overview of the concept of Natural Resource Accounting see Lutz and El Serafy (1988). For a discussion of the concept and an application to several sectors of the Indonesia economy see Repetto and others (1989). Readers interested in the economics of soil resources are referred to the works listed in the reference section especially Clark and others (1985), Magrath and Grosh (1985) and Sfeir-Younis (1985).  In Section IV the various costs of erosion are summarized and compared for the regions of Java. The Policy and methodological implications of the analysis are also considered. Although the analysis does result in an estimate of the cost to the economy of soil erosion it is important to acknowledge the limitations of the currently available information on soil erosion on Java. There are severe limits to the availability of reliable data on the rate at which soil erosion takes place on the different soils, the impact of this erosion on crops, farmer responses and all the other farm and nonfarm factors that determine the social losses caused by soil erosion. Nevertheless, the government of Indonesia as well as multilateral and bilateral donors have allocated millions of dollars to aid in the reduction of erosion. A major aim of the study is to illustrate a logically consistent framework in which the economic significance of a major form of environmental deterioration can be assessed. There remain important weaknesses in our understanding of the physical and behavioral processes that give rise to this deterioration and which make it socially and economically relevant. This paper illustrates the potential usefulness of scientific research on a number of issues in soil science. Until new and more definitive data become available, readers who are skeptical of particular assumptions are free to insert their own and explore their impact on our results. I. Measures of Land Degradation on Java Soil degradation is a gradual process that occurs as soil depth declines by erosion leaving progressively less topsoil and lower nutrient concentrations. Data monitoring this process on aggregate levels in Indonesia are not available. In this section alternative indicators of degradation are reviewed along with estimates of their quantitative significance on Java. The most frequently cited statistic on the severity of soil erosion in Java is the figure of 1.1 million hectares of "critical" land. Critical land is said to be increasing by 200,000 ha per year. According to the rough calculations of Ramsay and Muljadi (1983) land rehabilitated under the Regreening Programme, when adjusted for seedling survival rate, probably amounts to around 125,000 ha. On net, therefore, "critical" land area may be estimated as increasing by some 75,000 ha per year. More recent information from the Ministry of Forestry is that the total area of critical lands is now declining by 10,000 ha per year. Unfortunately, there does not seem to be a riF rous, generally accepted definition of critical lands. According to Roche (1987: 14) the old Indonesian Directorate of Landuse defined critical land only on the basis of slope. Any land with slope greater than 50% was designated critical. Obviously increments to critical land using this definition is nil. Ramsay and Muljadi (1983) cite the following criteria for critical lands: - unable to produce cassava yields of more than 500 kg/ha/yr2 2/ Presumably below this level cultivation is unattractive even for subsistence agriculture. Java-wide cassava yields are greater than 10 tons/ha.  . 3 - - unable to act as a regulator in the water system - unable to fulfill any protective function such as absorb run off. They further describe critical lands as areas in which all the topsoil has been removed by erosion and in which not more than 25 cm of subsoil remains in place over the parent rock material. While the amount of land in the critical category is a useful summary of overall seriousness of the soil degradation, it has several severe limitations. The accuracy and precision with which areas of land are designated as critical is open to considerable question. The extent to which the criteria listed above can be, or actually are, used in estimating critical areas is u,iknown. Discussions with GOI officials suggest that local authorities responsible for making these designations have considerable latitude. Consequently the de facto criteria for defining critical lands probably vary widely. A perhaps more serious problem with the critical lands concept is the fact that soil degradation is a gradual process that intensifies as soil depth decreases. The binary choice, critical/non-critical, provides only a poor approximation of the level or rate of change in the value of the aggregate soil resource. Long before soil quality falls to the point where agricultural production is completely unprofitable, there are discernible reductions in yields and net income. Only considering land that completely drops out of production will underestimate the severity of land degradation. Attempt to utilize available time series data on land use does not provide useful insights into changes in soil quality on Java. There are no time series on lands designated as critical (other than could be generated by using the rates reported above). In addition, the critical lands designation is relatively new. Due to these uncertainties, a modelling approach has been taken to estimating erosion levels in Java. In addition to providing greater consistency, this approach allows building on a large number of small area studies and allows for greater flexibility in designing components of the model to mesh with economic components. By making the relationships in the model explicit, it is also possible for users of the model to change variables to explore their consequences. II. The On-site Costs Of Soil Erosion A. Estimating the Physical Dimensions of Soil Erosion The physical dimensions of soil erosion that are of interest for this analysis are the areas of land affected by various levels of erosion and their spatial distribution. In order to generate this data an erosion model based on soil type and slope, land use, and patterns of rainfall intensity was developed. Soil erosion, as with so many other aspects of agriculture and land use, is highly location-specific. In addition, the random processes of nature can affect erosion rates over both time and space. Thus no mathematical model, inevitably limited to a relatively few independent variables, can replace  -4- detailed empirical measurements of erosion rates. Unfortunately there are currently no available data that provides an adequate empirically based picture of soil erosion on Java.3 Among the data that are available are maps at tha scale 1:1,000,000 of three variables that play a major role in determining erosion rates; i.e., soil types and slope, rainfall erosivity and land use. The soil map used for this study was published by FAO (1959) based on the work of Dudal, Supraptoharjo and others. It combines soil units with topography and is useful for visualizing the relations between soils and potential erosion. Twenty-five units are distinguished: - five units of soils on level to undulating land, with dominant slopes under 8% (units 01-05); - eleven units of soils on rolling to hilly land, with dominant slopes from 8-30% (units 06-16); and - nine units of soils on hilly to mountainous land, with dominant slopes over 30% (units 17-25). Descriptions of the soil types are included in Annex 1, areas of the soils by province are shown in Table 1. Rainfall erosivity, a measure of the kinetic energy released as raindrops strike the ground, is a major factor contributing to soil erosion, and its inclusion is essential in any assessment of erosion problems. Bols (1978, 1979) has prepared an isoerodent map of Java based on correlations of a measure of the kinetic energy of storms with annual rainfall data, which are available for most of Java over an extended time period. On Bols' map, eleven classes of rainfall erosivity are distiiguished at a scale of 1:1,000,000. Area estimates for each erosivity class are shown in Table 2. Land use data for Java is available in tabular form from the Central Bureau of Statistics; however, these data are of questionable reliability. Their iAost severe shortcoming is that they provide no means for correlating land use with the other factors that affect erosion. The Ministry of Forestry nonetheless produced a land use map of Java in 1985. On the basis of the map, the following four types of land use (or vegetation cover) have been distinguished for the objective of quantification of the actual erosion: - Areas of sawahs, including fish ponds. These areas are characterized by low erosion rates and, in fact, in large areas sedimentation prevails over erosion; Areas of Tegal (dryland farming), mostly on sloping uplands where erosion rates are very high; 3/ There are a number of small area studies that have been conducted on small watersheds and experimental plots. As explained below, these data, while not comprehensive enough to provide a complete basis for an economic estimate, are used throughout to provide a basis for the parameters in the model.  Table 1 SOILS ON JAVA (00 ha) Soil vest Central East Type Java Java Jogyakarta Java Java ,_._,_____............. .....____................. _"... _.___................... 1 86.7 485.1 0.8 213.0 785.6 2 9,203.7 5,514.1 409.6 9,286.5 24,413.9 3 614.0 189.7 52.7 180.5 1,036.9 4 359.8 2,550.6 30.9 2,101.3 5,042.6 5 0.0 271.2 0.0 0.0 271.2 6 0.0 0.0 219.8 535.2 755.0 7 1,714.4 3,482.2 181.3 2,012.2 7,390.1 8 454.3 0.0 0.0 136.6 590.9 9 112.0 1,635.5 0.0 1,457.8 3,205.3 10 6,464.1 3,946.0 0.0 1,902.2 12,312.3 11 666.9 0.0 0.0 0.0 666.9 12 2,232.2 99.1 0.0 42.6 2,373.9 13 1,737.7 0.0 0.0 0.0 1,737.7 14 3,081.4 364.1 817.9 3,835.4 8,098.8 15 0.0 1,941.9 29.5 1,528.2 3,499.6 16 0.0 0.0 0.0 265.9 265.9 17 1,161.2 1,753.2 602.0 2,598.2 6,114.6 18 0.0 71.0 0.0 4,259.0 4,330.0 19 0.0 0.0 0.0 117.7 117.7 20 5,372.1 3,352.3 0.0 3,592.3 12,316.7 21 1,495.1 1,854.1 0.0 2,268.4 5,617.6 22 5,811.9 0.0 0.0 0.0 5,811.9 23 7,688.7 1,311.4 0.0 0.0 9,000.1 24 240.4 1,988.9 339.5 1,903.5 4,472.3 25 1,603.4 860.5 518.9 5,028.4 8,011.2 TOTAL 50,100.0 31,670.9 3,202.9 43,264.9 128,238.7 S. ...u ...ce: Calculatd.fo.F. (5... Source: Calculated from PAO (1959).  -6- Table 2 AREAS OF JAVA SUBJECT TO ALTERATIVE LEVELS 0F EROSIVITY (00 ha) grosavtty Vest Central East Level Java Java Jogyakarta Java Java A 108.0 76.2 2.1 2,439.1 2,625.4 B 2,582.3 1,949.0 456.0 9,046.8 14,034.1 C 8,184.1 4,407.0 1,158.1 19,390.5 33,139.7 D 7,995.5 7,279.6 891.8 6,877.8 23,044.7 E 6,826.5 6,809.5 362.5 2,103.8 16,102.3 F 10,100.2 4,305.2 332.5 1,743.0 16,480.9 G 7,348.5 2,558.4 0.0 1,253.6 11,160.5 H 6,712.3 2,530.4 0.0 380.5 9,623.2 I 242.6 1,281.4 0.0 29.4 1,553.4 J 0.0 344.7 0.0 0.0 344.7 K 0.0 129.6 0.0 0.0 129.6 TOTAL 50,100.0 31,671.0 3,203.0 43,264.5 128,238.5 Source: Calculated from Boli (1978).  - 7 - - Forest areas, i.e., areas of natural and planted forest, including perennial plantation crops where erosion is slight; and - Degraded forest areas, including areas of shifting cultivation and degraded pekarangan (home gardens) where erosion is moderate to high. These four types of land use or vegetation cover are summarized in Table 3. Table 4 compares the areas of Sawah and Tegal as shown on the Ministry of Forestry Land Use Map with estimates published by the Central Bureau o4 Statistics (CBS). The Land Use Map estimates of Sawah area excsied the CBS estimates for every province, estimating almost twice as much Sawah area for Jogyakarta. On the whole of Java, the Forest Map estimates about one third more land in Sawah than the CBS. Forestry Ministry area estimates for Tegal range from 80 to 177 percent of CBS estimates. For West Java the Ministry of Forestry data exceeds that of CBS by 77 percent. For Java as a whole the divergence between the two sources is about 11 percent. The discrepancy over Sawah area is the most troubling. Unfortunately, there is no clear reason to prefer one source to the other. Sawah area is generally thought to be one of the more reliable statistics in the CBS land use data. However, the Forestry Ministry Map iL based, at least in part, on airphoto interpretation in which Sawah area is easily and accurately measured. Fortunately, because Tegal land is a more important source of soil loss than Sawah, the discrepancy is considerably smaller for Tegal with the exception of West Java. Given the Tegal area discrepancy, and because the cost of erosion calculation is performed first on a per hectare basis (see below) it was decided that it was most appropriate to use both sources. The Forestry Map was used in the soil erosion calculation only because it provided a spatial dimension that allowed correlation with the other elements of the soil erosion model. However, because the Central Bureau of Statistics data appear to be somewhat more reliable, these data are used in the final economic calculations. The three maps described above were digitized and analyzed using the Geobased System by the World Bank's Environmental Operations and Strategy Division.4 The procedure is essentially an overlaying of the three maps that identifies and provides an estimate of the areas of land characterized by the various combinations of slope and soil type, erosivity, and land use. Given the 25 soil groups, 11 erosivity classes and 4 land uses a total of 1,100 combinations are possible. In order to be able to take into consideration additional agronomic and economic differences, the analysis also divided Java along provincial boundaries resulting in 4400 possible combinations.5 Maps showing the distribution of the main soil types and slopes, and erosivity and land use are given in Annex 2. 4/ Rounding errors in the Geobased System program resu.t in minor discrepancies in area estimates. Consequently, columns and rows may not add exactly. The errors introduced in this way are insignificant. 5/ West Java, Central Java, D.I. Jogyakarta, and East Java, D.K.I. Jakarta was included in West Java.  - 8 - Table 3 LAND USE ON JAVA* (00 ha) ..... ............ ...... .................. ,. ............*. Land West Central Zast Use Java Java Jogyakarta Java Java ......... ......................... . .................... Sawah 16,043.8 13,361.3 1,078.5 16,969.6 47,453.2 Forest 5,412.7 6,357.6 0.0 12,075.1 23,845.4 Degraded Forest 3,009.2 340.3 31.5 615.2 3,996.2 Tegal 25,634.4 11,258.2 2,093.0 13,604.5 52,590.1 TOTAL 50,100.1 31,317.4 3,203.0 43,264.4 127,884.9 = ... ................... .. ... .. .................... ......... ........................... ........ Source: Calculated from Ministry of Forestry (1985). *Columns and rows may not add due to rounding. Wetlands excluded from analysis.  Table 4 COMPARISON OF IAND USE ESTIMATES (00 ha) ... ............... . . . ,... *,.,,,.................................... ., ..........,,,.......... CBS Ministry of Forestry Model As Percent ,............................. ............. ......................................... ...................... Sawah Area Estimates West Java4/ 12,152.74 16,043.8 132 Central Java 10,231.43 13,361.3 131 Jogyakarta 636.20 1,078.5 169 East Java 11,988.43 16,969.6 0 .............. ........................... ... JAVA 35,008.80 47,453.2 136 Tegal Area Estimates2/ West Java 14,402.14 25,464.4 177 Central Java 13,660.78 11,258.2 82 Jogyakarta 1,963.72 2,093.0 107 East Java 17,440.27 13,604.5 78 . ,,.4.......... ......... ... 47,466.91 52,420.1 110 Including D.K.I. Jakarta. House Compound and Surroundings and Bareland/Gorder/Shifting Cultivation.  - 10 - The estimate of actual erosion rates corresponding to each of the possible combinations is based on measurements under given conditions of plant cover or cropping, and on judgement based on erosion elsewhere under comparable conditions. Several recent projects on Java have yielded valuable data on actual erosion of uplands. These include the successive UNDP/FAO Projects in the Upper Solo watershed, the US-AID Project in the Citanduy watershed, Dutch sponsored projects in the upper Brantas (Kali Konto) and the Upland Agricultural Projects of Jogyakarta and the Jratunseluna and Brantas watersheds financed by U.S.-A.I.D... and the World Bank. Other erosion data have been collected by the Soils Department of the Agricultural University in Bogor, by the Soil Research Centre in Bogor and by the Watershed Management Centre in Solo. Still other erosion measurements have been reported in the literature from before the war and in other publications (see reference section). On the basis of these studies and observations, and based on judgement of local conditions, estimates of erosion rates of different soils under the influence of prevailing rainfall erosivity and under the major types of land use described above were calculated. The estimated levels of erosion resulting from this procedure need to be used with considerable caution. While they are believed adequate for the purpose of estimating erosion as an input to an estimate of the economic cost of erosion, the procedure is clearly not suited for other uses such as detailed land use planning. The model does not explicitly consider several important factors in determining erosion rates, particularly conservation practices and the considerable differences that can arise in ground cover within the broad categories of land use. For a discussion of the difficulties and pitfalls involved in even thoroughly tested erosion equations, see Wischmeier. Tables 5 to 8 give predictions for soil loss on the various soil types and land uses for the four regions of Java. Table 9 summarizes this data and shows that Tegal accounts for by far the greatest total amount of gross soil 1088.6 On a per hectare basis soil loss is highest on Tegal land on West Java, followed by Tegal on Central Java. The soils of East Java are least subject to erosion. Predicted soil loss on a per hectare basis is shown in Table 10. B. Estimating Productivity Effects of Erosion While it is widely accepted that erosion lowers agricultural productivity, there is little agreement on exactly how productivity is related to erosion or on the quantitative impact of erosion on yields. In part this results from the difficulty of defining fertility, as well as the difficulty of conducting controlled experiments to identify and measure erosion-related yield changes. Erosion involves changes in the availability and relative concentration of nutrients for plant growth, and changes in soil structure which influence root growth and affect the availability of water. Weathering of subsoil, which may be affected by soil management and by the roots of plants, may contribute some replacement of the factors that together constitute land 6/ This estimate, of course, only provides for soil erosion caused by rain and does not account for other sources of erosion such as mass washing and stream bank erosion.  - 11 - Table 5 PREDICTED SOIL LOSSES FROM TEGAL BY REGION AND SOIL TYPE (00 mt) Soil West Central East Type Java Java Jogyakarta Java Java 1 0.0 1,221.9 0.0 0.0 1,221.9 2 15,805.8 3,210.5 473.2 4,939.7 24,429.2 3 1,928.8 647.8 178.0 42.4 2,797.0 4 1,539.2 4,850.7 105.2 1,861.5 8,356.6 5 0.0 0.0 0.0 0.0 0.0 6 0.0 0.0 3,516.0 16,707.6 20,223.6 7 99,425.5 117,761.4 1,783.0 42,444.2 261,414.1 8 16,096.9 0.0 0.0 0.0 16,096.9 9 4,446.2 12,986.5 0.0 11,592.2 29,024.9 10 59,124.3 64,917.2 0.0 33,190.1 157,231.6 11 12,506.2 0.0 0.0 0.0 12,506.2 12 41,561.7 909.3 0.0 77.2 42,548.2 13 13,472.8 0.0 0.0 0.0 13,472.8 14 603,786.3 34,840.5 73,412.1 222,954.1 934,993.0 15 0.0 69,924.4 0.0 49,852.1 119,776.5 16 0.0 0.0 0.0 13,346.5 13,346.5 17 22,545.8 77,812.7 18,154.8 41,819.7 160,333.0 18 0.0 3,089.0 0.0 137,128.1 140,217.1 19 0.0 0.0 0.0 5,917.1 5,917.1 20 764,107.1 405,105.4 0.0 92,083.0 1,261,295.5 21 262,923.8 277,731.2 0.0 49,517.6 590,172.6 22 374,987.8 0.0 0.0 0.0 374,987.8 23 1,104,839.3 308,443.0 0.0 0.0 1,413,282.3 24 63,719.6 9,263.7 50,516.2 149,667.8 273,167.3 25 211,543.8 108,459.6 99,344.1 160,692.4 580,039.9 TOTAL: 3,674,360.9 1,501,174.8 247,482.6 1,033,833.3 6,456,851.6