Water as an economic input
Freshwater is a natural resource that provides many economic and environmental services (Briscoe 2005). Young and Haveman (1985) already noted 30 years ago that water has unique physical properties, complex economic characteristics, and important cultural features that distinguish it from all other resources. The idea that water resources management can benefit from economic principles can also be defended (Briscoe 2005) but should always take account of other, non-economic values that may restrict the scope of these principles.
The main sources of freshwater are rivers, lakes, and groundwater deposits (Zhu and van Ierland 2012). In our simulations, we focus on surface water that is supplied by rivers and through precipitation. We assume that groundwater deposits are not (further) depleted, so groundwater plays no role in the analysis. In the Netherlands, large-scale abstraction of surface water is subject to a license, but there is no tariff per unit of abstraction, so no water price as such. If freshwater becomes scarce (meaning that not all demand can be satisfied at present conditions), water gets a shadow price. The shadow price of water that is used by a certain activity is the value added that would be created by increasing the supply of water by one unit. Without water markets, shadow prices will, as a rule, differ across different economic activities and different locations. For this study, the term water markets is defined rather loosely as a mechanism which allocates water for economic use according to its shadow price and accordingly equalizes the marginal shadow costs of water use across economic activities. We do not describe the institutional setup of such markets, nor the physical infrastructure and associated investment costs that may need to be in place for water markets to function properly, nor do we take transaction costs into account. In addition, water markets in our analysis should be considered more as a yearly market for water use rights (for example, in the form of an auction) rather than as a spot market to satisfy immediate short-term water use needs.
The model
For our analysis, we use an extension of the well-known Global Trade Analysis Project (GTAP) model (Hertel 1999) called GTAP-Water (GTAP-W) (Calzadilla et al. 2010). GTAP is a comparative static CGE model of the global economy. GTAP-W extends the original GTAP model by adding an explicit treatment of irrigation water in crop agriculture and distinguishes between irrigated land and rainfed land.
The standard GTAP model (we use version 6) describes the interactions of decisions of consumers and producers in all markets. Consumers have preferences over private consumption goods, public goods, and savings, and they buy the consumption bundle that maximizes their utility, given their income, according to a constant difference elasticity (CDE) function. Producers maximize profits given a constant return to scale production technology for all firms. The competitive equilibrium in the model is characterized by clearance of all markets and by the zero-profit condition for all firms. The substitution between domestically produced and imported goods is imperfect, following the approach suggested by Armington (1969) to treat goods of different origin as different, non-homogeneous goods. The production function in GTAP is a nested constant elasticity substitution (CES) function. At the top level of the nest, intermediate inputs are combined with a value-added composite which consists of labor, capital, and land. The capital and labor endowments are perfectly mobile domestically, while land is imperfectly mobile. All endowments are immobile across regions.
GTAP-W extends the GTAP model by including more detail into the land endowment for agricultural producers (which are the exclusive users of the land endowment) splitting the original endowment into rainfed, irrigated, and pasture land and then further splitting off irrigation water from the irrigated land endowment. All of the new land and water endowments in GTAP-W inherit the partial mobility parameter from the original land endowment in GTAP. In the GTAP-W production function, in the lowest nest of the CES production function, crop farmers determine the level of irrigation based on the relative prices of land and irrigation water and the technical ease of varying the level of irrigation water on a given piece of land. In higher nests of the production function, the irrigated land is combined with capital, labor, and intermediate goods (seeds, fertilizers, pesticides) to produce an output like wheat or potatoes. Figure 1 provides a visual representation of the production function for the agricultural activities. The components of the agricultural water composite are discussed below.
We have extended the GTAP-W model in two ways.
First, we extend the existing GTAP-W model by including freshwater use in the animal husbandry sector, manufacturing sectors, and the public water services sector (which delivers drinking water to households and firms). In the animal husbandry sector, the demand for water is modeled in the same way as the demand for irrigation water in crop agriculture: water is combined with pasture land on the basis of relative prices and technical possibilities. In contrast, the manufacturing and public water services sectors combine water with capital (not land). The assumption of the possibility of substitution between water and capital is in line with the findings of Dupont and Renzetti (2001) and Renzetti (1992) who assert that intake water may be a substitute for recirculation water which is more capital intensive. The assumption is also in line with Gomez et al. (2004) and Goodman (2000) who combine water with capital in their models in a similar way. Solely for the purpose of the policy simulations that are carried out later in this paper, we distinguish between water from the agricultural water market that is combined with land in the agricultural sectors and water from the industry water market that is combined with capital in the manufacturing and public water services sector. Note that we also use the terms agricultural water market and industry water market rather loosely as described in Sect. 2.1. In the case when the sectoral water endowments are fixed and non-tradable, we use the terms to describe the aggregate sectoral water endowments used separately in the agricultural and industry sectors. Figure 2 provides a visual representation of the production function for the non-agricultural activities, and the components of the industry water composite are described below.
Second, we extended the model by accounting for volume flows of water between sectors. This addition is necessary to insure a physical water balance when water is exchanged between very different types of use. In GTAP-W as presented in Calzadilla et al. (2010), the agricultural water endowment (renamed here the agricultural water composite) represents all of the benefits (and also the expenditure) of irrigation for production (for detail see Sect. 2.3 below). This includes everything involved in irrigation, not only physical water but also irrigation equipment. We make the same assumption that the values of the agricultural and industrial water composite include not only the water itself but also the value of all of the necessary machinery for the water activity.
As the agricultural water endowment in GTAP-W from Calzadilla et al. (2010) is measured in millions of dollars, any redistribution of the endowment among agricultural sectors redistributes the value of irrigation without an explicit accounting of water volumes. If one assumes that the added value of a cubic meter of water is roughly the same for all agricultural sectors, then this is a reasonable structure for examining a market for water (or for water use rights) in agriculture. However, our paper examines water redistribution where the ratio between the value of the water activity in production (the value of the agriculture or industry water composite) and the volume of water involved in that activity can vary quite substantially between uses, and this needs to be accounted for to insure that the volume of water before reallocation is the same as the volume after reallocation (see Sect. 2.3, specifically Table 3 for the magnitude of this difference for the Netherlands).
To separate the value of physical water from the rest of the agricultural water endowment, we have changed the name of agricultural water ‘endowment’ to agricultural water ‘composite’, and split up it further into physical water volumes and dedicated agricultural water capital (see Fig. 1). Similarly, for the public water services sector and the manufacturing sectors, we first split off the industry water composite from the rest of the capital endowment where the industry water composite includes the value of all expenditures on water-related activities of abstraction, purification, use in production, and discharge. We then further split the industry water composite into the value of physical water volumes and dedicated industry water capital, which represents the value of water equipment (see Fig. 2). Dedicated water capital is immobile. Physical water volumes are mobile in principle, but the mobility is restricted in various policy scenarios to simulate the various water market alternatives. The physical water volume endowment is combined with dedicated water capital with Leontief production technology (no substitution allowed). The immobility of dedicated water capital and the Leontief production technology between dedicated water capital and physical water volumes means that the value of the water composite is completely determined by the amount of physical water available and that a percentage reduction in amount of available water for production results in the same percentage reduction in the water composite. The elasticity of substitution between the irrigated land endowment and the agricultural water composite is 0.05. The elasticity of substitution between capital and the industrial water composite for the manufacturing sectors is 0.5 and is 0.1 for public water services. The elasticities of substitution were calibrated such that a 10 % reduction in water availability would induce the same price water elasticities that are given in Rosegrant et al. (2002).
A sensitivity analysis of the results with respect to the new elasticities introduced into the model, of substitution between land and water for agriculture and between capital and water for manufacturing and public water services, has been performed. The main conclusions on the economy-wide impact of water and on the direction of the movement of water in response to the water market policy scenarios (see Table 5, Sect. 4) are insensitive to the specific values of the substitution parameters.
Data
We use the GTAP 6 data base which describes the world economy in 2001 and contains 87 regions and 57 sectors (Dimaranan 2006). We aggregate the regions and sector to focus on the water-using sectors in the Netherlands and its neighbors in the Rhine and Meuse river basin. The aggregated regions, sectors, and endowments used in this study are shown in Table 1.
Table 1 Aggregated regions, sectors, and endowments used in this study
To determine the share of the crop sectors’ use of irrigation water, irrigated land, and rainfed land, we used the procedure detailed in Calzadilla et al. (2011) and summarized as follows. First, the value of pasture land is simply the value of the original land endowment used by the animal husbandry sector. Then, for each crop sector, the remaining value of the original land endowment was split into irrigated land and rainfed land based on the volume of irrigated and rainfed production of the crop. Finally, the agricultural water composite was split off from the irrigated land endowment for each crop based on the ratio between rainfed and irrigated yields, where the value of the additional yield gain is attributed to irrigation water. We used data on irrigated and rainfed production and yields from the IMPACT model (IFPRI 2010) for all crop sectors and all regions except the Netherlands. For the Netherlands, we used data from a more detailed study of Dutch crop yields from Deltares (2013) because that seemed to be more accurate (for a detailed discussion, see Koopman et al. (2015)). To estimate the value of the agricultural water composite for the animal husbandry sector (which is split off from the value of pasture land), we used data on the volumes of water used for animal husbandry and also for irrigated crops. Then, we assumed that average value of a cubic meter of water in the animal husbandry sector is the same as in the crop sectors (see below for data sources on water volumes).
As far as we are aware, there are no studies which explicitly examine the value of water for manufacturing sectors in the Netherlands and Europe. However, there are a few studies on water use characteristics in manufacturing (see, for example, Reynaud (2003) for France and van der Zeijden et al. (2009) for the Netherlands). To determine the value of the water composite for the manufacturing sectors, we used a survey by Scharf et al. (2002) of Canadian manufacturers, which details expenditure on water extraction, treatment, recirculation, and discharge for several manufacturing sectors. We transferred these expenditures to our model regions by using the number of employees per manufacturing sector from Scharf et al. (2002) and Eurostat (Eurostat 2012) as a scaling factor. The value of the industry water composite for the Rest of the World (RoW) region (Table 1) was determined by imposing the same ratio of the value of the water composite to the value of capital as the rest of Europe region. Table 2 shows the total calculated expenditure on water abstraction, treatment, and discharge of water users in the manufacturing sectors in the Netherlands as well as the total output of each sector. For the value of the water composite in the public water services sector (not shown in Table 2), we relied on Teeples and Glyer (1987) who estimated a constant cost share of raw water of 18 %.
Table 2 Expenditures on water abstraction treatment and discharge for all manufacturing sectors
For volumes of water used in the manufacturing and public water services sectors, we used data from the Eurostat website for all regions in the study with the exception of the RoW. We used the Eurostat website (Eurostat 2014) for irrigated water volumes as well, with the exception of the Netherlands where we used data from Hoogewoud et al. (2013). The irrigated volumes from Hoogwoud et al. differ substantially from the Dutch irrigation data from the Eurostat website, but we use Hoogwoud et al. because it is consistent with Deltares (2013) which we use for the value of Dutch crop production and the direct effects of climate change on agriculture (Sect. 3).Footnote 1
For volumes of water used in the animal husbandry sector, we used the report by Ward and McKague (2007) on water consumed per head of various livestock types and then used the Eurostat website (Eurostat 2013) for the number of standardized livestock heads per region. The volumes of water used for the Rest of the World (RoW) region were estimated such that the RoW region had the same ratio of water to land, for the agricultural sectors, and water to capital, for the industrial sectors, as the Rest of Europe region.
In estimating the water volumes used by industry, we ignore water used for cooling, which takes place mostly in the energy sector, but also in certain types of manufacturing. Water used for cooling is a process which involves abstracting large volumes of surface water, but returning virtually the same quantity and quality, only slightly warmer. As water use for cooling involves very little consumptive use, it does not necessarily involve a trade-off between users. An exception is if the water temperature is already quite high, then warmed water could affect the ecosystem where it is discharged. Trade-offs between water for environmental and economic use and the resultant feedbacks of water quality on economic use are beyond the scope of this paper (see Brouwer et al. (2008), Dellink et al. (2011), and Zhu and van Ierland (2012) for studies that include water quality in assessments of water for economic use).
We were not able to obtain data on volumes of water used for individual crop and manufacturing sectors only on the total amount of water abstracted in manufacturing and separately for irrigation. We made the simplifying assumption that the average (shadow) value of a cubic meter of water is the same for all sectors within manufacturing and within irrigated crop categories. The value of the water use composite and the volumes of water used in each of these categories for the Netherlands are given in Table 3.
Table 3 The value of the water use composite and the volumes of water used in each of four use categories for the Netherlands in 2001
The volumes of water used and the associated value of the water endowments shown in Table 3 are for self-abstracted water only. All use of purified water purchased from third parties including all drinking water is incorporated into public water services sector. The Netherlands has a very extensive piped water network. The irrigated crops and animal husbandry sectors use quite a bit of drinking water. This may account for the rather low volumes of self-abstracted water used in the animal husbandry sector.
The average price per volume for drinking water deliveries charged to bulk water users by the Dutch public water utilities in 2001 is approximately 0.90 USD m−3 (Vewin 2002). Comparing this to the values given in Table 3, we see that the average value per cubic meter of self-abstracted water for agricultural users (0.32 USD m−3) is well below this price. However, the average value per cubic meter of self-abstracted water for the manufacturing sectors (3.75 USD m−3) is much higher than the drinking water price. This is due to the fact that most of the expenditure of manufacturing sectors on water use is on water treatment prior to discharge (Scharf et al. 2002).
In the Netherlands, there is no market for self-abstracted raw water; in fact, the government technically owns the water and allows firms to abstract it for their own use. Determining the exact value share of the water composite that should be attributed to the water volume endowment if the firms were to be granted property rights over the water that they use is beyond the scope of this paper. Therefore, to determine the value of the physical water volumes for each sector such that the markets clear in the benchmark equilibrium (or alternatively that the government has allocated water according to its shadow price), we use the procedure outlined in Appendix 2. The robustness of the results in the paper related to the assumptions made in Appendix 2 is given in Appendix 3.