The main purpose of the present work is to evaluate the implications of uncertainty in the gradual process of enhancements to the efficiency of irrigation water use, derived from a policy of modernization. This allows a better understanding both of the extent to which these improvements can mitigate the output losses derived from water constraints and the uncertainty effects in this mitigation related, for instance, to the timing of the introduction and take-up of new technologies. The possibility that such gains will never reach a level of advanced efficiency is also considered in line with recent studies that have shown that technology spillovers could not be observed, in spite of investment and the capacity to absorb innovations [1
A number of studies have highlighted the importance of dealing with uncertainty in both technological change and climate change ([2
]). Uncertainty is a key issue for the design of climate change policy. The timing of the introduction of new technologies and the take-up of those technologies is difficult to estimate and is, therefore, uncertain. A decisive role in water policy is played by technological improvements. Prior studies have seen both risk and uncertainty in, for example, the implementation process of agricultural technology innovation for farmers ([6
]). Moreover, that uncertainty also extends to the availability of water resources. A review of papers dealing with uncertainty in different areas, such as climate damage, technological change, and emissions policies, can be found in [4
], which discusses the various conclusions of a range of studies and highlights the different lines of research. More recently, Berck, P. et.al [8
] presented a framework to make better decisions when facing different sources of threats, differentiating uncertainty as endogenous when it is created by human action or exogenous if it is created by nature, and address both risks at the same time. In this context, this work mainly addresses uncertainty as internal risk based on uncertainty in the gradual process of enhancements to the efficiency of irrigation water use by farmers through an irrigation modernization process.
To this end, our work stems from a prior study of technological change in irrigated agriculture, taking the largest irrigation scheme in the Ebro River Basin (Spain) as a case study (see [9
]). The region is semi-arid, suffers from irregular and significant water shortages, and is in the midst of a modernization process of irrigated land to switch from flood irrigation to sprinkler and drip irrigation systems. The water supply in the Ebro River Basin is irregular and there are serious doubts about its future trends. However, some studies have recently identified a cyclical and decreasing evolution in water flows that lets us assume a forecast framework of water supply based on the observed evolution [10
For the purpose of this paper, we use a recursive dynamic computable general equilibrium (CGE) model that includes elements that allow us to consider different alternatives for its evolution simultaneously and involve all variables of the general equilibrium model (prices, outputs, taxes, demands,…) although assuming perfect foresight of water supply into the distant future for each alternative. CGE models are empirical versions of a Walrasian model, formalized by [11
] and based on a set of numerical equations that capture the characteristics and general functioning of an economy. These models can measure both direct and indirect effects of alternative economic policies, as well as changes in the behaviour of economic agents. CGE models let us take into account the global impact in all sectors and agents after a shock in a specific sector, including price and rebound effects of technological improvements, which cannot be analyzed with partial equilibrium models.
CGE models have been widely used in recent years to address the impacts of climate change and greenhouse gas (GHG) emissions ([12
]) and also to address the management of water resources ([9
]). As these models consider the various linkages between economic sectors, they are particularly useful for the evaluation of water-pricing policies, due to the ability of these models to calculate both equilibrium prices and quantities [25
]. Such models have become a standard tool in the quantitative analysis of policy intervention in many domains, including fiscal policy, trade policy, and environmental policy [26
]. Additionally, these models have been extended to consider risk and uncertainty. Goodman, D.J. et. al [27
] presented a review of computable general equilibrium models that consider risk and quantify the monetary value of risk by incorporating uncertainty by formulating different paths in a CGE model. Harrison, G.W. et. al [28
] use stochastic structures to analyze the optimal policy mix between taxing emissions and subsidizing technologies in Research and Development programs, taking into account the date on which the advanced technology became available. The stochastic programming includes a set of possible states that is included in each variable of the model, and each possible state involves a different probability. They find that policy recommendations are different when uncertainty in technical change is modeled than when it is not. Moreover, as [4
] point out, these results demonstrate that the implications of uncertainty cannot always be only inferred from the standard sensitivity analysis of the model parameters. Heal, G. et. al [29
] show how uncertainty about the 2020–2050 emissions targets may affect CO2
and energy prices, as well as technological choices in the energy sector. To do this, they develop stochastic policy scenarios within a CGE framework to analyze CO2
Therefore, the main contribution of this work is to use a recursive dynamic CGE model, in order to address the effects of uncertainty in the gradual process of enhancements to the efficiency of irrigation water, focusing on the uncertainty in the timing of the introduction and take-up of new technologies. As far as we are aware, very few studies have so far incorporated uncertainty in CGE models in this area (see [27
], among others). Pervasive uncertainty, the likely revision of policies over time, and unexpected changes in technology, suggest the need for a recursive-dynamic solution in the CGE analysis, as [2
] conclude. Solving a complex CGE model was, until recently, difficult due to our limited computational resources, but the use of such complex models has become possible following advances in mathematical programming and software, which have facilitated the design of structures rooted in economic reality to reflect modeller objectives and the common sense of policy-making, in order to arrive at viable conclusions and to assess the implications of a wide spectrum of possible results.
This paper is organized as follows. Section 2
explains the extension of the model and presents the types of uncertainties tackled. Results are shown in Section 3
. Finally, Section 4
presents our conclusions.
We show the results of the main variable affected by water availability (output from irrigated agriculture) in Scenarios 1, 2 and 3. Although we assumed that water supply fluctuates, for simplicity, we present the average results for the first period (2002–2010), the second period (2011–2019), the third period (2020–2028), and the fourth period (2029-2037) in Table 1
. We also present the results in the largest drought years (2005, 2014, 2023, and 2032) in Table 2
. Additionally, we show results in irrigated agriculture prices in Table S2 of the SI
. In Table 1
and Table 2
, we can also see the results of comparing our benchmark scenario
with a benchmark scenario
without declining water supply, which is the standard steady state of an economy—variables and resources grow at a constant rate. We consider that these last figures allow us to observe the impacts of declining water supply.
The upper part of both tables shows a comparison with our “non-modernization” benchmark scenario with the different technological alternatives and the expected value for each scenario, assuming that all of them have a declining water supply. Thus, we can observe the improvements achieved for each possible alternative and the expected improvement of each scenario. The lower part of the table reports the range of variation of the different evolutions of technological change and the influence of the uncertainty associated with the occurrence of each alternative and each scenario, considering as a reference the current expected evolution in the area (Huesca province, Spain) to reach the expected level of efficiency (80%) by 2030. For this reason, we compare all scenarios with the alternative ‘2030’ of Scenario 2, which could represent the current expected evolution based on data on the current modernization process in CGRAA.
Firstly, in the left part in both tables, we can see negative results when we compare with the benchmark scenario without declining water supply due to the strong fall-off in the water supply during the period 2002–2040. These negative figures are a good reference measure of the effects produced by the water restrictions shown in Figure 4
. However, in the three scenarios in both tables, we can also see that the improvement in efficiency in the use of water, through a policy of modernization, reverses, partially at least, the negative effect associated with the declining water supply. However, we can see that the relative capacity to mitigate negative impacts is larger in the most recent periods (compare the impacts of declining water supply with the rest of the scenarios), showing that the mitigation of negative impacts of declining water supply will be more difficult to revert in the future years because of larger expected water constraints. See, for instance, in Scenario 1 with alternative ‘2020’, in the period 2002–2010, we obtain a difference of 5.97 percentage points between modernization and non-modernization, which equates to around 70% of the 8.52 of losses after the declining water supply. While in the period 2029–2037, the difference of 16.61 percentage points in alternative ‘2020’ is around 42% of the 39.62 of losses obtained with the declining water supply.
also reflects the influence of the speed of the advance of the modernization process. We observe that gains from technological modernization are bigger when technological progress is achieved sooner, rather than later. For instance, there is an improvement in the output of irrigated agriculture of 3.97 percentage points difference in the first period (reaching an advanced technological level in 2030), 9.36 in the second period, 12.69 in the third period, and 14.40 in the fourth period, in Scenario 1. Similarly, these gains are greater when technological advances are faster; for instance, 12.39 percentage-points difference in the period 2011–2019 with a fast process, ‘2020’, compared to 7.51 with a slow process, ‘2040’. Note that a very slow process without reaching the 80% efficiency level until 2100, known as ‘2100’ in Scenario 3, presents hardly any gains (percentage-points differences) over the periods and is closer to the scenario without modernization with small benefits.
More interestingly, the lower part of Table 1
and Table 2
clearly shows the differences in the timing of the introduction and the take-up of new technologies by comparing with the expected evolution of the alternative ‘2030’ of Scenario 2. The positive effects associated with the higher technology options, which achieve 80% efficiency from 2020 to 2030, are greater than the gains from the technologies that achieve it from 2030 to 2040, because we assume a Gompertz evolution for technological change. The effect is particularly noticeable when advanced technology with the modernization process is attained in ‘2020’ and ‘2025’. Note that these results confirm the relevance of taking uncertainty into account in the timing in the adoption of technology in an irrigation modernization process.
As the results in Table 1
are presented as average values of the period, Table 2
allows us to observe results in the largest drought years. The negative effects induced by water constraints in our non-modernization benchmark scenario
are larger than in Table 1
, and again improvements are obtained after implementing the modernization process (see also the lower part of Table 2
). However, it is more difficult to reduce output falls in drought years in Table 1
. Therefore, similar findings are obtained in Table 1
and Table 2
, but we can add that the consequences of the take-up of new technologies to mitigate the impacts of water constraints are more drastic in drought years. The faster and earlier the take-up of technologies, the greater will be the reductions of agricultural output losses.
Focusing on each alternative (date on which the top efficiency is achieved) in each scenario, we can see that the type of uncertainty—“high uncertainty” in Scenario 1, “normal distribution” in Scenario 2 and “uncertainty including failure” in Scenario 3—involves different expected results for each scenario. Particularly, expected values show larger values in Scenario 1 due to the certainty that there will be technological progress. Scenario 3 clearly represents the worst situation due to the greater probability of failure in the modernization process with low expected values. However, we can see that when the occurrence date alternative is the same—for instance, the results in alternative ‘2030’ in all scenarios—there are very small differences between scenarios.
In order to analyze the influence of the type of uncertainty, as probabilities affect prices, we show the impacts on irrigated agriculture prices in Table SI2 of the SI
. We can see that a declining water supply involves price increases due to water scarcity. The policy of modernization achieves increases in productivity that lead to a reduction in the price increase without modernization. This is especially clear when we look at the values of expected prices. Moreover, the reduction in the price increase is greater if advanced technology is achieved by 2020, than if it is achieved later.
Additionally, we observe that if the expected value of achieving a top level of efficiency is higher (Scenario 1 versus Scenario 3), the reduction in price increase obtained with modernization is slightly larger in the farthest periods. This slightly larger reduction in price increase (i.e., lower prices) provokes slightly lower benefits in irrigated agriculture production; see Table 1
. Then, the type of uncertainty in the probability of the event does not have a relevant role in the results associated with each alternative, and results, for each isolated alternative, are mainly defined by technology.
To sum up, the inclusion of uncertainty in the evolution and adoption of the irrigation modernization process allows us to observe the effectiveness of the policy of modernization under different conditions and the importance of its timing, because the benefits of technological improvements could be reduced or even lost due to a delay in the implementation of advanced technology. Indeed, efforts to achieve technological improvement in the use of water could be wasted, depending on the date on which the top efficiency is achieved, and technology improvements have hardly any effect in drought years. In other words, the date on which the top efficiency is achieved is a key element in any policy of modernization in the agriculture sector.
4. Concluding Remarks
The inclusion of simultaneous sensitivity analyses to address the implications of uncertainty in the technology of irrigation water management, in a computable general equilibrium model, highlights certain conclusions. Specifically, this study evaluates the implications of uncertainty in the gradual process of enhancements to the efficiency of irrigation water use—as much by the timing of the introduction and take-up of new technologies as by the uncertainty associated with the occurrence of each alternative.
We find that a policy of irrigation modernization through improvement in water use always mitigates the negative impacts associated with water constraints, including the price increases due to water scarcity. However, profits from improved technology could be reduced or even missed if the implementation of the advanced technology is delayed. The date on which advanced technology becomes available, is adopted, and enters general use, is quite significant. In other words, a fast technology take-up is important to mitigate the effects of declining water supply. However, mitigation of the impacts of water constraints could be more difficult to alleviate in severe drought years. Negative results can be reduced with technology, to a greater or lesser extent, depending on the date on which the top efficiency is achieved.
Therefore, greater efforts, involving large investment by farmers and society, are required to ensure the benefits of a policy of modernization. It should be noted that the uncertainty in the timing of the introduction and take-up of advanced technologies can be affected by the impact of additional uncertainties in the evolution of water supply as an external risk. In line with the analysis provided by Poudel and Paudel (2018), an increase in the external risk of larger downward water supply, together with the internal uncertainty in technology, should lead to an increase in precautions. Then, farmers and planners could mitigate the expected negative results through the modernization process with the adoption of different measures such as the use of testing, scientific studies, precision farming, seed optimization, irrigation and property management, systems of crop monitoring, etc., that allow the gradual assessment of the modernization process.
Another conclusion from this study is that the external inclusion of simultaneous sensitivity analysis with probabilistic options in CGE models reveals additional findings that cannot be obtained directly from a deterministic CGE model. Additionally, in this first study on the evaluation of these kinds of uncertainty, we also study different probability assumptions presented in three scenarios. The findings show that the type of uncertainty in the probability of the event does not have a relevant role in the results when the occurrence of the date alternative is the same. However, the expected results of each scenario are different and can change significantly in accordance with the farmer’s expectations (alternative probabilities). Undoubtedly, further research on the impacts on other dynamic CGE models, with different specifications of the agents’ expectations, is required to provide a better comprehension of uncertainty. As a consequence, an extension of the present study requires both a deep analysis of agent (farmers, politician, …) decisions and the exploration of other sources of uncertainty such as the evolution and character of water supply. CGE models can be powerful and useful tools if they continue to be updated and adapted to the outstanding issues.