1998 AAEA paper
A Computable General Equilibrium Approach to Surface Water Reallocation Policy in Rural Nevada
Chang Seung* Thomas Harris Rangesan Narayanan
Selected Paper at 1998 American Agricultural Economics Association Meeting Salt Lake City, Utah August 2-5, 1998
_____________________________________ * Chang K. Seung is a Post-doctoral Research Fellow in the Department of Applied Economics and Statistics at University of Nevada, Reno. Thomas R. Harris is a Professor in the Department of Applied Economics and Statistics at University of Nevada, Reno. Rangesan Narayanan is Chair and Professor in the Department of Applied Economics and Statistics at the University of Nevada, Reno.
Send correspondence to: Chang Kyu Seung, Dept. of Applied Economics and Statistics, Mail Stop 204, University of Nevada, Reno, Nevada 89557-0105. Phone: (702)784-1356; Fax: (702)784-6701; E-mail: [email protected]
A Computable General Equilibrium Approach to Surface Water Reallocation Policy in Rural Nevada
Abstract This study uses a computable general equilibrium model to examine the impacts of transferring water from agriculture to recreational use in rural Nevada. Model results show that different assumptions about input substitution in agricultural production produce qualitatively different policy impacts on agricultural sectors.
Introduction This study examines the economic impacts of reallocating water from irrigated agriculture to recreational use at the Stillwater National Wildlife Refuge (NWR) in Churchill County, Nevada. This study uses a standard regional computable general equilibrium (CGE) model. This study focuses on how sensitive the agricultural impacts of a water reallocation policy are to alternative possibilities of input substitution in agricultural production. The study region, Churchill County, Nevada, contains both the agricultural land from which water will be withdrawn and the Stillwater NWR wetlands to which the water will be reallocated. Years of drought during the late 1980s and early 1990s degraded the quality and quantity of water reaching the refuge. This severely impaired fish, upland game and migratory waterfowl habitat -- an important consideration since approximately 30% of migratory waterfowl in the Western United States feed and rest at Stillwater. In addition to supporting numerous species, Stillwater also provides recreational opportunities such as bird watching, waterfowl hunting and camping. Estimated visitation at Stillwater has ranged from 28,000 to 40,000 visits annually. Visitors are estimated to spend more than $1.1 million each year which translates into an additional $440,000 of direct and indirect income to Churchill County (Loomis, 1985). There exists a rich body of literature, including numerous state and regional economic impact studies of water management in western states, which address the tradeoffs from alternative water policies. A classic example of state-specific impact study
is Seckler (1971). Other examples include the evaluation of irrigation development and hydroelectric power generation in the Pacific Northwest (Hamilton et al., 1982). More recent studies include Dinar and Zilberman’s (1991) model of the interaction between agricultural drainage salinity and economic impacts in the San Joaquin Valley of California and Berck et al.’s (1991) investigation of water reallocation in the South San Joaquin Valley, California. Recently, Leones et al. (1997) studied the economic impacts of recreation in the Rio Grande River Basin near Taos, New Mexico. Transferring water from agriculture to the wetlands at Stillwater will reduce agricultural production within the study area and increase water-related recreation activities and expenditures both within and outside the study area. This study focuses on how sensitive the agricultural impacts of a water reallocation policy are to alternative possibilities of input substitution in agricultural production. We consider four model variants, each of which represents a different possibility of factor substitution in agricultural production. Model results show that different assumptions about input substitution in agricultural production result in qualitatively different policy impacts on agricultural sectors. However, this study shows that for some model variants, the model results for agricultural sectors are more or less similar.
Model Description Production.
There are eight production sectors. Three of them are agricultural
sectors which include (i) livestock, (ii) other crops, and (iii) hay and pasture. The other five sectors are non-agricultural sectors which include (iv) mining, (v) construction, manufacturing, transportation, communication, and public utilities (CMTCPU), (vi) trade, (vii) finance, insurance, and real estate (FIRE), and (viii) services. Production technology in each sector is represented by a Cobb-Douglas (CD) value added function. A constant returns to scale technology is assumed for each sector’s production. Intermediate inputs are used in fixed ratios. Agricultural sectors use labor, capital, water, and land as primary production inputs. In this study, it is assumed that crop substitution does not occur when water is withdrawn from agriculture. This assumption is reasonable for agriculture in the study area where land characteristics limit the ability to switch crops. Therefore, we assume that a fixed amount of water is combined with a unit of land in agricultural sectors, and that withdrawal of a certain amount of water from agricultural sectors implies a proportional reduction of land use in those sectors. Non-agricultural sectors use only labor and capital as primary factors of production. Profit-maximization for each sector’s production yields its factor demand for each factor of production. Factors of Production. Supply of labor has two sources---labor from inside of the region and labor from outside of the region, i.e., labor in-migration. It is assumed that labor is mobile across sectors such that sectoral distribution ratios of wage rates are
maintained. It is assumed that labor is partially mobile between regions depending on the interregional differential in the average wage rates. We consider four model variants, each of which represents a different possibility of factor substitution in agricultural production. In each of the model variants, land use in each agricultural sector is assumed to be reduced by about 51.2 percent due to the policy of withdrawing water from agriculture. Each model variant is described in the table below.
Model 1 Model 2 Model 3 Model 4
Description Both labor and capital are reduced proportionally with reduction of land use due to the policy. In this model variant, no factor substitution is allowed in agricultural sectors. Capital use in the agricultural sectors are proportionally reduced when land use is reduced due to the policy shock. Only labor is substitutable for land and capital. Both labor and capital in each agricultural sector are allowed to vary in response to the policy shock, and are substitutable for land. Capital in each agricultural sector is fixed at its base-year level both before policy and after policy. Therefore, only labor is substitutable for land in agricultural sectors.
Consumption. Following IMPLAN, households are grouped into three types, i.e., low income households, medium income households, and high income households. Preferences of the households are represented by a constant elasticity of substitution (CES) utility function. Each type of household is assumed to consume locally produced goods and imported goods from outside of Churchill County. Utility maximization for each type of household subject to its budget constraint yields its demand function for each good. When water is removed from the agricultural sectors, farmers are compensated based on their water rights. Some of the farmers who receive compensation may continue
to reside in Churchill County, while others may move out of the region. Expenditures by the farmers who remain in the area will change, while expenditures for those farmers who leave Churchill County will be lost. Additional information is required to accurately assess the number of farmers who will remain or leave Churchill County and to calculate the net change in expenditure levels. Without such information, this study can not model the migration/expenditure behaviors of the farmers who receive compensation for their water rights. Therefore, in this study, the net change in farmers’ expenditures is assumed to be zero.
Empirical Implementation In this section, model results from the CGE model will be discussed when there is 125,027 acre-feet of water inflow to the Stillwater National Wildlife Refuge wetlands. This amount of inflow will be met with the acquisition of 101,000 acre-feet of water rights from agricultural producers and drainage from remaining irrigated acreage (MacDiarmid 1988). The reduction in agricultural land use due to withdrawing water from agriculture is about 51.2 percent compared to the base year. Data. IMPLAN database for 1992 was used to make a SAM for Churchill County, Nevada. The 528 sectors in the Churchill SAM were aggregated into the eight sectors in this study. Elasticity values used in the CGE model are from previous econometric studies. To calibrate the CGE model, non-elasticity parameters were solved for given base-year values of the model variables, values of elasticities, and the particular
functional forms for the model equations. Incomes of the three types of households were designated by IMPLAN software where low income households earn less than $20,000; medium income households earn between $20,000 and $40,000; and high income households earn more than $40,000. To calculate base-year factor income in each of the agricultural sectors, we used factor shares developed by Robinson et al. (1990). For nonagricultural sectors, we treat employee compensation as labor income and the combined proprietors’ income and other property income as capital income. For data on water and acreage available in agricultural sectors and at the wetlands after the water transfer, we used the information from MacDiarmid (1988). The author found that with the acquisition of 101,000 acre-feet of water rights from agricultural production, the water available in the agricultural sectors would decline from 197,280 acre-feet to 96,280 acre-feet. The inflow to the wetland would increase from 88,945 acrefeet to 125,027 acre-feet due to the water transfer. Details are found in Table 1 or in MacDiarmid (1988). With the reallocation of the agricultural water to the wetlands, tourism expenditures by recreation visitors would increase because the surface area of the wetlands would increase. Using a general population mail survey of Nevada residents, Harris et al. (1998) estimated the relation between the numbers of trips by angling, general recreation, and hunting visitors and the water supply to the Stillwater National Wildlife Refuge wetlands using a seemingly unrelated regression method. The authors found that
the size of water acreage at Stillwater National Wildlife Refuge has a positive influence on number of recreators in hunting and angling. Details are found in Harris et al. (1998). Also from the general population survey of Nevada residents, expenditure patterns by Nevada residents who traveled to the Stillwater Wildlife Area were derived by Harris et al. (1998). The authors found that per trip expenditure made in Churchill County is estimated to be $11.60 for gasoline, food, and supplies and $12.50 for lodging. In our study, the expenditures for gasoline, food, and supplies are allocated to the trade sector and the expenditure for lodging is allocated to the services sector. Thus, with the increase in trip activity from 125,027 acre-feet of water inflow to the wetlands, trade sector expenditures made within Churchill County increase by $1,740 while services sector expenditures made within Churchill County increase by $1,875. These increases in expenditures are very small compared to the size of the Churchill County economy; the total increase in expenditure on the recreation-related sectors of $3,615 is only 0.0006 percent of the base-year value of aggregate production in Churchill County economy. Therefore, the increases in the recreation-related expenditures are expected to generate negligible economic impacts in our CGE model. Analysis of the Model Results. Given the decreased land use in agricultural production due to withdrawal of water and the increased expenditures for the trade sector and the services sector, the net economic impacts from transferring surface water from irrigated agriculture to water-related recreation are analyzed using the CGE model described above. We calculated the net economic impacts using each of the four model
variants when there is 125,027 acre-feet of water inflow to the Stillwater National Wildlife Refuge wetlands. Comparing model results from models 1, 2, and 3 (Table 2), it is seen that the policy impacts on agricultural output are smaller if the degree of factor substitution in agricultural sectors is higher. Model 1 reports the largest impacts on the agricultural output. This is explained by the assumption in Model 1 that both labor and capital are proportionally reduced when land use is reduced due to water withdrawal. Model 4 reports the smallest impacts on agricultural output—about 38.2 % reduction compared to the base year. This is because Model 4 assumes that capital in each agricultural sector is fixed at its base-year level. Similar results are observed for agricultural employment and agricultural labor and land income. The policy impacts on agricultural capital income are largest in Model 4, which reports a dramatic decline of about 44 percent in rental price of capital in agricultural sectors. Overall, Model 4 reports the smallest agricultural impacts while Models 1, 2, and 3 report more or less similar results. Policy impacts on non-agricultural sectors are very small for all variants of the model. In all model variants, the output and employment in the nonagricultural sectors increase slightly. This is because the labor and capital released from the agricultural sectors flow into the non-agricultural sectors, increasing output and employment of in those sectors. The largest increase in non-agricultural output and employment is reported by Model 1 because more of labor and capital is released from the agricultural sectors, and is absorbed in the non-agricultural sectors in Model 1 than in any other model variants.
Conclusion In this study, we employed a county-level CGE model to evaluate the impacts of reallocating water from agriculture to recreational use in Stillwater National Wildlife Refuge wetlands in Churchill County, Nevada. This study focused on how sensitive the agricultural impacts of a water reallocation policy are to alternative possibilities of input substitution in agricultural production. Model results show that different assumptions about input substitution in agricultural production result in qualitatively different policy impacts on agricultural sectors. However, unless capital in each agricultural sector is fixed at its base-year level, model results for agricultural sectors are more or less similar (Models 1, 2 , and 3). Several research directions are in order. First, this study focused on estimating only the intra-regional economic impacts of a water reallocation policy, ignoring the extraregional impacts of the policy. For example, this study did not estimate the change in recreation-related expenditures made outside of the study region and the competitive impacts of expanded recreation in the study region on other recreational areas in Nevada. A multiregional CGE model could be developed to capture these extra-regional effects. Second, due to lack of information regarding the expenditure/migration behavior of the farmers who receive compensation for their water rights, this study could not incorporate them into this study. Further research needs to identify the behavior of farmers who receive compensation. For example, one can obtain the information through a survey on
how much of the farmers’ compensation the farmers will use for paying debt and how much of it they will use for consumption of goods. Once the behavior of the farmers is identified, it can be incorporated into our model to determine how the model results will differ. Third, this study estimated only the economic impacts of a water reallocation policy, ignoring the change in consumer welfare from increased recreation activities. For a more complete benefit-cost analysis, the CGE model used in this paper needs to incorporate a framework for measuring welfare change from a change in recreation activities.
Water and Acreage Available in Agriculture and Wetlands Base Condition
350,636 (acre feet)
350,636 (acre feet)
Transportation Loss (B)
153,356 (acre feet)
153,356 (acre feet)
0 (acre feet)
101,000 (acre feet)
Farm Delivery (D = A-B-C)
197,280 (acre feet)
96,280 (acre feet)
Farm Delivery Per Acre (E)
3.7 (acre feet)
3.7 (acre feet)
Water Rights Acquisition (C)
Irrigated Acreage (F = D÷E)
Wetlands Transfer Rate (G) a
0 (acre feet)
81,619 (acre feet)
Drainage to Wetlands (I)
88,945 (acre feet)
43,408 (acre feet)
Total Wetlands Inflow (J = H+I)
88,945 (acre feet)
125,027 (acre feet)
Wetlands Delivery (H = G×C)
Wetlands transfer rate of 0.81 is calculated as the use rate of 2.99 acre-feet per acre in agriculture divided by the farm delivery of 3.7 acre-feet per acre.
Impacts of Water Reallocation in Churchill County, Nevada. Model 1
Output Agriculture Non-agriculture
Employment Agriculture Non-Agriculture
Labor Income Agriculture Non-agriculture
Capital Income Agriculture Non-agriculture
Land Income Agriculture
Berck, P., S. Robinson and G. Goldman. 1991. “The Use of Computable General Equilibrium Models to Assess Water Policies.” In A. Dinar and D. Zilberman, eds. The Economics and Management of Water and Drainage in Agriculture. Norwell, MA: Kluwer Academic Publishing Dinar, A. and D. Zilberman (Editors). 1991. The Economics and Management of Water and Drainage in Agriculture. Norwell, MA: Kluwer Academic Publishing Hamilton, J., G. Barranco and D. Walker. 1982. “The Effect of Electricity Prices, Lift, and Distance on Irrigation Development in Idaho.” American Journal of Agricultural Economics 64:280-285. Harris, T., J. Englin, R. Narayanan, T. MacDiarmid, S. Stoddard, K. McArthur, and M. Reid. 1998. “Distributed Impacts of Surface Water Reallocation Policies.” Unpublished manuscript under review. Leones, J., B. Colby, D. Cory and L. Ryan. 1997. “Measuring Regional Economic Impacts of Streamflow Depletions.” Water Resources Research 33: 831-838. Loomis, L. 1985. Newlands Project Area Recreation and Socioeconomic Manuscript on File. URS Company, Sacramento, California. MacDiarmid, T. 1988. An Economic Analysis of the Efficiency Target Policy for the Carson Diversion on the Newlands Project. Unpublished Thesis, University of Nevada, Reno. Robinson, S., M. Kilkenny, and K. Hanson. 1990. “The USDA/ERS Computable General Equilibrium (CGE) Model of the United States.” USDA/ERS Staff Report No. AGES 9049 Seckler, D. (Editor). 1971. California Water: A Study in Resource Management. University of California Press, Berkeley, California.