Beyond Water Stress: Structural Adjustment and Macroeconomic Consequences of the Emerging Water Scarcity

Abstract

This work analyzes some system-wide macroeconomic consequences of lower (sustainable) water availability , when global economic growth is postulated according to the Shared Socio-Economic Pathway 1 (SSP1), for the reference year 2050. After finding that the rather optimistic forecasts of economic development cannot be met in most water scarce macro-regions, we assess what consequences for the structure of the economy, welfare and the terms of trade, the insufficiency of water resources would imply. The analysis is undertaken by means of numerical simulations with a global computable general equilibrium model, under a set of alternative hypotheses. In particular, we consider whether (or not) the regional economic systems have a differentiated capability of adaptation (by means of innovation and modification of economic processes), and whether (or not) the scarce water resources can be allocated among industries, such that more water is assigned where its economic value is greater.

Appendices

Appendix 1: An Assessment of Future Water Deficits

Although this study focuses only, as an illustrative case, on the year 2050 and the SSP1 scenario, future potential water deficits were estimated for other years and another scenarios (see Roson and Damania 2017).

The assessment of the future water deficits has been based on a limited set of SSP forecasts of income and population growth, complemented by CGE simulations aimed at enlarging the number of estimated economic variables. For each combination of year and SSP, growth rates in population and GDP have been assumed, using data from the IIASA SSP repository. By shocking the corresponding parameters in the GTAP CGE model (dataset 9.0), several other endogenous variables were obtained, like production volumes by industry and region, household consumption, regional investments, exports and imports, income by source, etc.

Estimates of industrial output are especially relevant because, coupled with some econometrically computed future water intensity coefficients, allow to derive the implied water demand. Analogously, municipal water demand was computed by assuming it dependent on population growth, real income levels and a trend of increased water efficiency . Table 4.7 presents the water demand projections for the SSP1 at the year 2050.

Table 4.7 Water demand projections (potential demand consistent with SSP scenario 1)

Regional water deficits are defined as the difference between potential water demand and sustainable water supply. In turn, the latter is identified as the sum of water runoff and inflow in a region, estimated by the global hydrologic GCAM model,³ driven by three different Global Circulation Models (CCSM, GISS, FIO ESM). We found that four macro-regions have levels of potential SSP1 demand exceeding sustainable supply in the year 2050.

Appendix 2: Estimation of the Marginal Value and Output Elasticity of Water

When water is regarded as a production factor, the Marginal Value of Water (MVW) is the increase in the value of output potentially obtainable when one unit of water (here, one cubic meter) is added to the process, while keeping all the other production factors unchanged. The concept is strictly linked to that of water pricing and allocation: (a) profit maximization and cost minimization imply that MVW should equate the price of water; (b) water (or any other resource) is efficiently allocated (from an economic viewpoint) when its marginal value is the same across alternative uses.

In principle, estimating the MVW would require specific technical information on the production processes and how water contributes to them. This is simply impossible to get for large aggregate sectors and regions. Instead, we propose here a methodology for a consistent estimation of MVW in 15 industries and 14 macro-regions, based on some available “water intensity coefficients” (WIC—water per value of output) and two calibrated parameters. WIC (indicated in the following as ω) and MVW are related but distinct concepts. Mathematically, WIC is just the ratio of water over output (in value terms), whereas MVW is the partial derivative of output value with respect to water.

The estimation procedure is based on a set of sensible assumptions one could impose on the water elasticity of output (ε). The latter is defined as the relative (percentage) variation of output (x) obtainable through a relative variation in the water input (w), ceteris paribus:

$$\varepsilon = \frac{\delta x/x}{\delta w/w} = \frac{\delta x}{\delta w}\frac{w}{x} = \frac{\delta x}{\delta w}\omega = MVW\omega$$

(2)

Consider ε to be a function of ω. Obviously, one would require that ε(0) = 0, because no variation in output would be observed if water is not used at all. A second sensible assumption is:

$$\mathop {\lim }\limits_{\omega \to \infty } \varepsilon \left( \omega \right) = 1$$

(3)

Meaning that, as water becomes the only relevant factor (enormous amounts of water are employed), the output varies proportionally with water (constant returns to scale). A smooth function with the two properties above would then be characterized by ε′(ω)> 0 and ε″(ω) < 0: the marginal value is positive but decreasing.

One of the simplest mathematical functions that can be adopted to express ε(ω) is the powered semi-logistic one:

$$\varepsilon \left( \omega \right) = \left( {\frac{\alpha \omega }{1 + \alpha \omega }} \right)^{\beta } \omega \ge 0 \; \beta > 0$$

(4)

By plugging (3) into (1), and solving for the MVW, a relationship linking MVW to WIC (ω) is obtained:

$$MVW = \omega^{ - 1} \left( {\frac{\alpha \omega }{1 + \alpha \omega }} \right)^{\beta }$$

(5)

This allows us to infer the marginal value of water on the basis of the water intensity, once the values of a and β have been set. We calibrated the values for these parameters using some estimates by Moolman et al. (2006), who computed the MVW for five categories of fruits in South Africa, in the year 2002, and our own estimates of the industrial water intensity for the year 2004 (Roson and Damania 2017). The beta parameter is calibrated by imposing that MVW equals 1.312 (simple mathematical average of the estimates by Moolman et al. cit.) when WIC (ω) is 0.01039 (our estimated value for Vegetables and Fruits in South Africa). The alpha parameter is simultaneously obtained through numerical optimization, imposing the requirement that the variance of MVW values by Moolman et al. equals the variance of MVW across South-African industries (excluding the outlier Services). The computed values are 0.637 for alpha, 0.855 for beta. Table 4.8 presents the corresponding MVWs.

Table 4.8 Industrial MVW 2004 (US$/M³)

Notice that, the higher the average productivity of water (value of output per m³, the inverse of the WIC), the higher the marginal value of water. In this respect, allocating water resources on the basis of the relative industrial water productivity (as it is done in the DRES scenario) is conceptually equivalent to allocating water on the basis of the relative marginal values.

The output elasticity of water is the percentage increase in gross production volumes obtained through higher water utilization. If no adjustment takes place in the production processes and in the use of other factors, then the elasticity is just the product of MVW and WIC. To get more meaningful effects when water availability is varied, we allow in this study some implicit adjustment in complementary factors, by expressing the output elasticity of water (η) as a linear function of the product (ε):

$$\eta = \gamma + \delta \varepsilon$$

(6)

where the γ and δ parameter values are set so that the average elasticity is 0.8 and the standard deviation is 0.2. Table 4.9 shows the elasticities obtained in this way.

Table 4.9 Industrial output elasticity of water

Appendix 3: The Construction of Regionally Differentiated Impact Scenarios

Simulations under the DRES and DRUS scenarios are based on the assumptions that regions, in which potential demand for water exceeds sustainable supply, differ in their capability of absorbing the excess demand (water deficit). The absorption percentages applied in the various cases are based on a mixed qualitative-quantitative analysis of the relevant characteristics, where we keep distinct the potential of technological innovation from the degree of flexibility in the economic structure and trade flows.

Looking first at the innovation side, notice that a number of technologies and management options can be put in place to improve the water efficiency (lowering demand) and/or expanding the water supply. Theoretically, the different options could be ranked in terms of economic efficiency, from the lowest to the highest unit cost, and those whose unit cost (possibly including externalities) falls below the shadow value of water (increasing as the water gets scarcer) should be selected (WRG 2009). In practice, however, the technological response to the water stress is much more complicated, as a variety of factors (technical, political, institutional, safety, etc.) ultimately affects the choice among the different technology options (Becker et al. 2010).

We therefore rely on a qualitative index of technology potential for each of the potentially water stressed macro-regions, based on a subjective evaluation of several options and characteristics. Because of the subjective and qualitative nature of this index, the latter should be interpreted as expressing an informed scenario, rather than as a solid scientific appraisal of (future) technical capability in the regions.

We consider three important classes of technology or management options:

• 1. Desalination

• 2. Enhanced irrigation techniques and reduced evaporation

• 3. Water reuse.

For each of them, we identify five “facilitating factors”, possibly making the implementation of each option more likely:

• 1. Physical conditions (e.g., desalination projects will be more effective if most of the urban centres are found along the coast).

• 2. Factor availability (e.g., access to energy sources for desalination ).

• 3. Institutional capacity (efficient level of government, quality of public institutions).

• 4. Human and physical capital (relevant for large and complex projects).

• 5. Demand potential (e.g., enhanced irrigation is primarily targeted to agriculture, therefore its effectiveness depends on the share of agricultural water on total water consumption ).

We assign to each factor in each region and for all the three alternatives above a simple scoring system: 1 (poor), 2 (average), 3 (good). A “Technology Potential Index” (Table 4.10) is quite naturally obtained by simply adding up all the given points. The higher this index, the easier is the expected capability of a region to adjust to water deficits through the introduction of new technologies and more efficient management techniques.

Table 4.10 Regional technology potential index

A second adjustment mechanism is related to the endogenous changes in the regional economic structure. Indeed, when actual water availability turns out to be lower than what would be required for production and consumption purposes, the consumers’ utility diminishes and the productivity in water-using industries declines. Even in the absence of a formal market for water resources, scarcity is transmitted as a price signal, and a structural adjustment takes place in the economic system, alleviating the overall impact of the negative shock for the economy. What is maybe less known is that the same process leads to an improvement in the aggregate water efficiency or productivity (water per unit of output), whose magnitude—however—depends on a series of specific characteristics of the economic system under consideration.

Many factors contribute in determining the structural flexibility , and it is not easy to ascertain what economies could respond better and why. To shed some light on this issue, we performed a simple numerical experiment with the global general equilibrium model. In each of the potentially water stressed macro-regions, we simulated a −10% reduction in multi-factor productivity in agriculture, which is the sector where most of the water is utilized. The consequent drop in total agricultural output volume is shown in Table 4.11.

Table 4.11 Agricultural output change

A CGE model cannot capture all the factors and characteristics affecting the actual degree of flexibility in a certain economy. Nonetheless, a simple experiment like the one above can offer an order of magnitude, or at least can suggest a ranking of the regional economies from the most rigid one (Central Asia) to the most flexible one (East Asia), in terms of absorption of productivity shocks in agriculture, possibly induced by water scarcity .

We combine the ranking provided by Tables 4.10 and 4.11 to split the absorption of the excess water demand in the three components: internal structural adjustment, technical and management solutions, and reduction in water delivery. The latter component, which is obtained as a residual, determines the amount of decrease in water delivery (with effects on productivity) in the scenarios DRUS and DRES (Table 4.12).

Table 4.12 Decomposition of excess water demand absorption