This paper examines the water allocation system and its effects on water use, users' behavior and on the country's achievement of desirable goals. It investigates efficiency, equity, and sustainability issues that arise under this allocation rule. The current allocation system functions according to a queuing system of priorities, where the urban allocation is found to be the most efficient and least risky allocation, while the ecological sector bears the highest degree of climate risk variability. The rule of priority is applied to crops as well; the non-strategic crops have the least efficient and riskiest allocation. The current system allows the country to secure drinking water supply, but it does not create sufficient incentives for entitlement holders that have priority to increase their water use efficiency, does not guarantee ecosystem health and integrity and does not equally distribute the risk among users. An institutional reform is especially relevant to improve water use performance in the agriculture sector and the country's ability to manage drought. The nonpriority system that allows farmers to exchange their water-use entitlement might increase social welfare of water use.
In Tunisia, the Water Act defines water as a public good that is managed by the public water authority. The system of water rights is based on water access rights that are fundamentally simple usage rights determined by bureaucratic prescription in a public allocation system (Republic of Tunisia, 1975). The water resources are managed according to a hierarchical approach and the water planning authority makes decisions centrally. The Ministry of Agriculture is designated as the water authority, which plans water development, decides on allocations, and coordinates and controls the interests; its regional departments execute what is decided, while at the local level, water user associations manage water in irrigated lands, but they do not participate in the decision-making process. The National Water Utility (henceforth SONEDE), which is under the control of the Ministry of Agriculture, holds the monopoly on drinking water supply over the entire country, otherwise some water user associations manage drinking water for some rural populations.
Tunisia is a semi-arid country that experiences a situation of absolute water scarcity as defined by Falkenmark (1989) and is prone to the persistent risk of intermittent droughts (Louati et al., 1999). With fast-expanding water needs and growing concerns about the increased scarcity, improving water-use performance and drought management capacity are challenging further the country that has nearly reached the boundaries of conventional water development.
This paper attempts a critical study of water resources management. Water allocation effectiveness is a fundamental parameter by which the performance of water management can be evaluated. The water allocation institutions or system is the set of rules that are used to determine and to share the available water resources among different users. The Tunisian water allocation system and its performance implications have not been analyzed before. To our knowledge, there are no studies to date to help decision-makers, managers, and users understand the effect of current allocation institutions on water use.
While the Water Act, the strategies and the resources development master plans are officially recorded in water-management government files, none of them clearly defines water allocation schemes. In the Tunisian context, water allocation rules are not perfectly visible for all interest groups and water users, they are not officially established but are de facto rules developed within a centralized decision-making process. Actually, the water allocation plans are rather subjected to power-sharing rules and policies, while being influenced by agricultural and regional development factors. Therefore, we attempt first to describe and explore the features of the Tunisian water allocation system, then we investigate its performance implications. This analysis should provide decision-makers with useful information to design more appropriate regulatory arrangements in order to improve water management in Tunisia.
The understanding of contextual conditions, which are highly relevant to water management analysis, requires a comprehensive research strategy. Case studies will be used to address how and why the water allocation institution is designed and set up. Case studies are the ‘problemsheds’ of water management, as they are where the principles, actors, and situation of water management converge (Grigg, 1996).
Tunisian water allocation mechanism
Document review, meetings observation, data examination and interviews allow gathering of factual information on the water allocation system and related decision-making process. The meetings attended took place in the Ministry of Agriculture, in which representatives of its central and regional departments came together to address issues of concern relating to the water allocation plan. The office of water planning and hydraulic equilibriums of the Ministry of Agriculture, which is in charge of preparing the water allocation plan, provided access to meetings that were otherwise closed to the public. The review of water law and water resources development strategies gave some insights on water resources management objectives and principles. Interviews with actors who participated in setting the water allocation plan and with stakeholders affected by the allocation decision were relevant to gain an understanding of perspectives involved in water management. In addition, the database of the Ministry of Agriculture was used to reconstruct historic water allocation in the context of case studies.
Design and practice of water allocation rule
The design of the water allocation plan is aligned with the farming schedule. At the beginning of the winter season (i.e., September/October), the water authority develops the water allocation plan for the season as well as a provisional plan for the summer season. This plan is implemented through a quantity-based administrative allocation and it specifies how the quantity of water that is expected to be available for use during the upcoming season will be shared among uses. The water sharing is subject to technical constraints, such as the possibility of transfer, the purpose of the dams, the quality of water, and the purpose of water use. The development of the allocation plan depends on the quantity of water that the water authority decides to put into use and on the quantity adequately preserved for future use during an expected drier period. In the context of the semi-aridity of Tunisia, the water authority deals with the fulfillment of current demand and protecting the country against future drought conditions.
In practice and in order to develop the water allocation plan, the water authority estimates the volume of water that will be available for use in each hydrosystem based on the volume impounded by the dam at the beginning of the season, and on climatic and hydrologic forecasts for the upcoming water year. The water authority also collects, from regional departments of the Ministry of Agriculture and from the SONEDE, regional agriculture water requirements and drinking-water requirements, respectively. The regional departments assemble, from water user associations, the water requirements of the farmers' cropping plan for the upcoming season. Based on these data, the central administration prepares the water allocation plan that specifies the allocations by type of use and by region and the source of water that will provide the allocated quantities. The central administration discusses therefore, with the regional departments, the technical modalities to execute that plan, i.e., the area of irrigation, the cropping plan, the handling of irrigation and the source that will provide the water quotas. Subsequently, the regional departments inform the local water users associations about the water allocations decisions. The farmers adapt their cropping plans accordingly as the water authority defines water quotas in terms of irrigated surface and type of crops and often imposes water rationing on farmers' demands. Hence, the irrigators' cropping plans should be approved by the regional department that provides water to irrigators, with respect to the seasonal contract of irrigation established between the regional department and the water users associations.
In Tunisia, surface water resources represent 56% of the total country's available water resources, 78% of which are located in the north of the country. Tunisia has mobilized almost 90% of its exploitable water resources, half of which are stored in reservoirs (Tunisian Ministry of Agriculture, 2015). The water allocation process concerns the water impounded by dams. Therefore, the case studies selected for this analysis were located in the north of Tunisia (Figure 1). In addition, this region accounts for approximately 80% of Tunisia's agricultural activities.
The Ichkeul watershed is a sub-drainage basin of the Extreme North and Ichkeul watershed. It was the primary case chosen because the government recognizes a right to water for the Ichkeul ecosystem, which is a unique case in Tunisia. In addition, Ichkeul dams are part of the water transfer network that is designed to supply the country's coastal areas with water taken from North and Extreme North areas. Therefore this case study might illustrate the competition over water between uses that may arise at multi-region scale.
The second case study focuses mainly on the irrigated agriculture sector. Agriculture uses approximately 80% of the country's mobilized water resources, more than half of which are surface water resources. The irrigated lands, which do not exceed 8.2% of the usable agricultural area, account for 37% of production value. The Buherthma, which is a sub-drainage basin of the Mejerda watershed, has been selected from a list of other candidates because the main purpose of Buherthma dam is to provide local farmers with water, and therefore this case could be viewed free from regional and cross-sector competition.
Case study 1: Ichkeul watershed
Ichkeul watershed is a humid climatic zone in the Bizerte region, located in the north-eastern part of the country. Although its area covers no more than 1.31% of the country's total surface, it provides a regular water supply that represents 12.40% of the country's surface water resources (Tunisian Ministry of Agriculture, 1998). The Ichkeul watershed is characterized by an ecosystem that is a Ramsar site and a United Nations Educational, Scientific and Cultural Organization (UNESCO) world-heritage site. Since the 1980s, the construction of dams on three among six rivers of the watershed, has deprived the lake of a substantial quantity of its freshwater natural inflow, causing disruptions to the ecological conditions of the ecosystem. Part of the water stored in Ichkeul reservoirs is transferred to be used outside the watershed as these dams are designed to serve at least two main purposes: drinking water supply and drought management according to the hydrological situation at the country level. As a result, the ecological needs of Ichkeul ecosystem, in terms of fresh water, come up against competition from uses in the Capital, Cap Bon and Sahel regions of the country.
The Ichkeul water allocation data, which are provided by the Ministry of Agriculture, contain information on dams' releases and dams' water uses. Actually, the releases from Ichkeul dams (i.e., Sejnane, Joumine and Ghazala dams) put into effect the water allocation plan. Therefore, this information has been used to reconstruct the water allocations practices, and the allocations of Ichkeul water dams are aggregated into four categories of use, in spite of the fact that the uses take place inside or outside Ichkeul watershed: urban, agriculture, Ichkeul, and environment use. The Ichkeul use refers only to the volume of water supplied specifically to the Ichkeul lake, whereas the environmental use is the sum of the volume for groundwater recharge and the Ichkeul use in addition to releases related to dam operations, which still may have some ecological benefits for Ichkeul lake. The urban use is the volume of water allocated to the SONEDE.
The water allocation strategy is based on the rule of assigned priorities to each water use. In practice, this rule is implemented during deficit years. In wet years, the demands are not rationed, generally. Water allocations during the period of study are displayed by use in Figure 2. The red dots (resp. blue dots) depict the deficient year allocations (resp. non-deficient year allocations) of pairs of sectors. In any year, if the allocation for the use along the horizontal axis is larger than the allocation for the use along the vertical axis, the dot will be below the 45-degree line.
We note that half of the period of study, 1990–2012, has been faced with a hydrologic deficit (Figure 2). Drought is a frequent phenomenon in Tunisia, even if its severity has a significant inter-annual variability (Louati et al., 1999). During the deficient period, the Ichkeul water allocation is the smallest in comparison with those of other uses (Figure 2(b) and 2(d)), and environmental water allocation is still smaller than the urban and agriculture allocation in most cases (Figure 2(c) and 2(e)). For wet years, however, it is clear that much more water goes to the Ichkeul Lake, but it is also likely that all of the agricultural and urban needs are satisfied.
In order to explore the effective water allocation rule, we construct the empirical distribution of water allocation (henceforth eCDF) of urban, agricultural and environmental uses.
Figure 3 illustrates a descending order of water allocation reliability by use. The urban allocation seems to be more reliable than the agricultural one which is more reliable than the environmental allocation, i.e., for a given benchmark quantity of 20 106 m3, the probability of the urban user's delivery exceeding this benchmark quantity is around 89%, while this probability is about 54% and 14.5% for farmers and environment, respectively. In a more realistic world, however, each user has a specific requirement, for example, let us suppose that urban, agriculture and environment ask for 20 106 m3, 30 106 m3 and 25 106 m3 respectively. Then urban users have 89% chance of receiving their claims, while farmers and environment have, respectively, 24% and 6% chance of obtaining the quantity requested. This differential degree of reliability is the implementation of the order of priority assigned to uses during the water allocation process.
Case study 2: Buherthma basin
The Buherthma region in the north-western part of the country, in the Governorate of Jendouba, is situated in the Upper Medjerda Valley. Traditionally, dry cereal crops are grown in the Buherthma region. However, for the sake of improving the land's yield, in 1976, the Tunisian Government began a dam irrigation project across the Buherthma River and implemented a sprinkle irrigation system to supply a 24,000 ha area, which accounts for almost 6% of the country's irrigated land. The project was designed to promote food security by increasing the country's production of sugar, milk, cow meat, commercial and home garden crops and wheat to meet food sufficiency needs, something that weighs heavily on the country's balance of payments. Moreover, the project aims at reducing poverty in rural areas by increasing farmers' revenues and creating jobs (African Development Bank, 2006).
The project introduced a major innovation by encouraging the growing of sugar beets, resulting in the construction of a sugar refinery. However, the factory was closed some years later, mainly because sugar beet cultivation was found not to be cost-effective for farmers in the region.
The Buherthma dam is designed to supply the local irrigation demand. The sprinkle irrigation system is fed by two sources: the Buherthma dam and the Medjerda River, whose low water discharge is augmented by water from the Mellegue dam, situated south-west of Bousalem.
The data used to describe water allocation in the Buherthma region were retrieved directly by the author from annual activities reports, which were requested from the Jandouba Regional Department. The data display information on areas by crops and water billed in Buherthma irrigated land. The required data on Buherthma dam's releases were provided by the Ministry of Agriculture as well.
At the beginning of the farming season (i.e., September), the water authority prepares the water allocation plan for the region based on the water stored in Buherthma dam and on the hydrological forecast for the upcoming season. If the water year is anticipated as deficient, the water authority designs the allocation plan for crops growing in the region, which are classified into strategic crops (i.e., cereal, feed plants and potatoes) and non-strategic crops (i.e., market garden, leguminous, industrial market garden). Then the water authority informs water association users, through its regional department, of the allocation plan. Therefore farmers, who are organized as water association users, adapt their crops growing plan and conclude a seasonal irrigation contract with the regional department.
Based on data collected, we have reconstituted the history of water allocation for strategic and non-strategic crops categories. Figure 4 depicts the empirical distribution (eCDF) of allocation to strategic crops and non-strategic crops respectively.
During the period of study, strategic crop uses have benefited from more reliable water allocation than non-strategic crops (Figure 4). Then the hierarchy of use, which is used to allocate water in agriculture, is put into practice by assigning a different degree of supply reliability to uses, and by granting a most secure delivery to strategic crop uses.
Water allocation pattern
Three aspects guide the characterization of a water allocation system: how available water is shared between various uses and between users, how risks and burdens associated with water use and management are distributed among members of the community, and how responsibilities for causing environmental harm are distributed between stakeholders. Allocation is closely intertwined with access, as rules and procedures of water allocation shape the right to access to water. The distribution of access rights could be seen as the fourth aspect of the allocation system, which cannot be completely depicted without examining how the access rights are distributed (Gupta & Lebel, 2010; Doorn, 2013).
The allocation of surface water described in the previous section can be formulated as follows.
In the context of the water allocation decision-making process, the water authority deals with natural random water inflows and, consequently, with random reservoir annual yields. The decision-maker adopts specific reservoir operation rules to meet predefined objectives. Let X represent the total random release from a reservoir in a given season. In other words, X represents the total amount of water available for use in a given season. X is a continuous random variable with a probability density function (i.e., the probability distribution of release) that is strictly positive in [0,d] and zero outside the interval and where d is the maximum release in a given season.
Water authority plans to allocate the available quantity of water at a given season among users that are categorized by the type of their demands. Let k denote the number of water demand types. Under this condition, X can be partitioned into k random variables Xi that is the water allocations to demand type i. Then, is also a continuous random variable with a cumulative distribution function (henceforth CDF), , which is increasing in [0,d].
represents the water authority allocation plan, which could vary from season to season according to water resources variability, caused by climate fluctuations, and according to the water allocation rule. During a period of water shortage, the public decision-maker favors certain demand types over others. He considers, in fact, that a particular water use has a higher social benefit than another. The water authority defines the rank of priority to every water demand type and designs water entitlement for each of them according to their assigned priority rank. Each type of entitlement provides a volume of water at a specific degree of reliability. In this context, the reliability of water allocation is defined as the probability, that the volume supplied in a given period is sufficient to fully serve demand (Alaouze, 1989). The degree of reliability of supply for a specific demand is correlated positively to a rank of priority assigned to this demand. Therefore, the water allocation rule grants to every type of demand an allocation at a specific level of reliability.
Formally, let and be the allocation to the demand type s and type j respectively. We suppose that the demand type s has seniority in term of priority over the demand of type j. The reliability of water allocation to the demand type s and type j, are measured by and , respectively. The allocation rule previously described implies that . Consequently, the cumulative distribution of (i.e., the probability distribution of allocation facing the junior user type that has a low rank of priority) lies entirely or partly above the cumulative distribution of (i.e., the distribution facing the senior user type). In that case, is set to be stochastically larger than , in other words, dominates in the sense of first-degree domination (Fishburn, 1977; Kamae et al., 1977). Then the probability of the senior's allocation exceeding benchmark quantity x is always greater than the probability of the junior's allocation exceeding that benchmark quantity. This relationship holds for any given benchmark quantity and for any junior-senior comparison. If x denotes the minimum quantity desired by a user in a given season, then the probability to be served at least with x is lower for the junior user type than for any senior user type.
The stochastic ordering seems to be an attractive approach to analyze the current water allocation system as it is a nonparametric approach and therefore it does not impose restrictions on the functional forms of the probability distribution of water allocation. Bawa (1982) and Mosler & Scarsini (1993) review and give detailed bibliographies on the theory of stochastic dominance and its many theoretical and empirical extensions and the wide spectrum of application areas, including economics and finance applications.
From a practical point of view, the stochastic dominance criteria require a pairwise comparison of cumulative distribution functions. In application, however, the compared distributions are usually unknown and must be inferred from the observations. As the empirical distribution function converges to the distribution function (Tucker, 1959), the stochastic dominance rules are commonly applied to empirical cumulative distribution functions derived from data describing the action. The empirical distribution may serve for water users as estimates of future distributions of water allocation (Levy, 2015).
The case studies performed in the previous section show that the dominance relationship in the sense of the first order dominance (FSD) can be established between all pairs of allocation series, i.e., (urban, agriculture), (agriculture, environment), (urban, environment) and (strategic crops, non-strategic crops). In fact, between any two distributions, one empirical distribution function (eCDF) lies entirely above the other and no pair of eCDF ever crosses (Figures 3 and 4). Moreover, the FSD algorithm conditions hold. A FSD algorithm serves to check empirically whether the FSD relationship is effective or not, i.e., and there is at least one strict inequality. Where the observations and are recorded from the lowest to the highest value (Levy & Hanoch, 1970). Then we can assert that urban water allocation dominates agriculture water allocation that dominates environmental allocation, and strategic crops allocation dominates non-strategic crops allocation.
So far, the analysis has focused on examining the water allocation rule as a whole. This institutional arrangement can also be examined from the perspective of water users. Quirk & Saposnik (1962) and Hadar & Russell (1969) relate the stochastic order relationship with the actors' preferences order: where denotes the expected utility and is the set of monotonic increasing utility functions. In the context of our analysis, this theorem can be interpreted as follows: the entitlement to the demand type s is unanimously preferred to the entitlement to the demand type j, by all water users that have a monotone increasing utility function, if and only if . Thus, an expected utility-maximizing user would rank entitlements in the same order as if the ranking were based on the degree of reliability attached to the entitlement. The reliability of the supply is the most relevant criteria from the point of view of water users.
The stochastic dominance approach is used here also to order water users' preferences without imposing explicit specifications for these preferences toward water entitlement. The utility function is supposed only to assume a natural preference order, that is, water users prefer to get more water than less, as they derive benefits or welfare from using it. This assumption seems plausible in the context of a water allocation under scarcity conditions and where there are competitions on water use.
Then, from the point of view of water users, the dominant entitlement is the most efficient. If, in addition, water users are risk-averse, the dominant entitlement is the less risky one (Hadar & Russell, 1969).
Under this water allocation mechanism, unequal sharing of risk and access to the resource among water users are implicit in the hierarchy of use purpose. The senior claimants obtain a preferred position due to their priority in access to the resource and junior users bear a high degree of climatic variability risk and the burden of the shortage falls completely on them. They endure also the risk that the water authority manages the hydraulic system or the drought episode sub-optimally.
What are the performance implications of the current water allocation system?
The answer to this question might give insight into appropriate institutional changes to improve water resources management; however, evaluating performance is not trivial. The performance implications of the water allocation rule can be measured by the degree of well-being that society can reach by using its water resources under specific conditions. The increase in benefits from water use, in relation to cost that is a performance improvement, should enhance well-being. Therefore, the evaluation of water benefit and cost values is the key to assess the performance implications of the water allocation rule. From an ecological-economics approach perspective, Lant (2004) has classified water benefit values into market, economic and ecological economic value categories and arranged water costs in supply, economic and ecological-economics cost classes. Market value is a portion of economic value, which is, in turn, a portion of the ecological-economics value. On the cost side, the supply cost is a part of the economic cost, which is a part of the ecological-economics cost. The best use of water is then the ecological-economics use with the greatest net value that leads to the highest social well-being.
The definitions and measures of benefits and costs are therefore multiple as water is a resource with multifaceted uses at different stages of its cycle and provides a wide range of short, medium and long-term benefits, both for humans and for ecosystems. Syme et al. (2008) have depicted in ‘Sphere of Needs’ individual and community needs that should be met by water use. Accordingly, needs have been classified into utilitarian needs (i.e., survival and health, wealth, prestige and social identity) and humanitarian needs (i.e., social cohesion, recreation, aesthetics, moral and cultural, spiritual). Evaluating benefits derived from water may become elusive especially for water uses that do not carry direct market value. Water has an economic value, but water is not an ordinary economic commodity, in the sense that conventional market instruments cannot determine every water benefit value (Savenije, 2002; Hanemann, 2005). Batten (2007) noted that only the part of the total water cycle nearest to the intermediate and final users could be recognized as an economic good. Valuation methods commonly used fail to provide sufficient avenues for articulation and assertion of many values that lie outside market and private-property institutional frameworks (Batten, 2007; Morehouse, 2011). The neoclassical market-based tools cannot reflect, in particular, critical information about the state and the quality of the resource and about processes that sustain its ecological services (Straton, 2006).
Wateau (2011) has analyzed various societies' modes of operating in using and distributing water among community members. She showed that local communities adopt different rationales to distribute water, such as social hierarchy, kinship, genuine lineage, and seniority. Water is used also to reaffirm identity, it is seen as a symbol and heritage. Based on various anthropological examples, Wateau brings evidence of some non-market values of water and highlights that social rationale could be stronger than economic and technical rationales during the water allocation process.
Therefore, the water value-in-use (i.e., the benefit derived from water) is rather a socially constructed one, so that each society applies principles that should guide the water allocation decision-making, in order to achieve desired outcomes. These values and principles emerge from actors' interactions within the social-ecological system. The values, principles and desired outcomes are revealed through formal rules, rule-in-use, informal rules, and discourses. The value of water and the conception of guiding principles are not immutable as they could be promoted and contested by society. Thus, the examination of societal processes of consent and actions allows decryption of the socially desired outcomes.
A blend of positive and normative approaches is therefore required to assess objectively the performance effect of the water allocation arrangement. The positive approach is the answer to the question ‘does the water allocation arrangement ensure that the society realizes its desired outcomes?’ We evaluate here the gap between desired goals and realized outcomes. The normative analysis, on the other hand, is the answer to the question ‘what ought to be an optimal water allocation?’ We analyze here whether the current allocation leads to realizing the ideal goals that the water resources management strategy should fulfill. The desirable goals are defined by the water management paradigm adopted. The water paradigm is an underlying pattern of thinking behind the water resources management approach shared by a community at a given time and an agreed method regarding how problems are perceived, how issues are addressed and how responses should be examined. It is a set of basic assumptions about the nature of the system to be managed, the goals of managing the system and the ways in which these goals can be achieved (Kuhn, 1996).
According to the water paradigm adopted, predefined normative criteria illuminate the relative strengths and weaknesses of the current water management strategy and guide towards attaining desirable goals. In the present study, the assessment of the performance implications of the current water allocation system is viewed through the sustainable water management approach prism, where efficiency, equity, and sustainability are the criteria commonly used to evaluate public policy (Daly, 1992; Roa-García, 2014).
Based on interview data and official discourse, the water allocation strategy has been driven by the necessity to secure drinking water supply, ensure food security and mitigate drought effects.
Regarding the first goal, Tunisia has achieved high access rates to drinking water service. In 2012, 97% of the population had access to improved sources of water (World Health Organization/United Nations Children's Fund Joint Monitoring Programme (WHO/UNICEF JMP), 2014) and 83% of the population had a property connection with 24 out of 24 h service supply. In practice, the urban water priority is a priority assigned to the demand of Tunisian Water Utility that has held the monopoly on drinking water supply over the entire country since 1968. In spite of the high rate of coverage reached, the performance of the drinking water industry has become slack over the last decade, resulting in a suboptimal use of water resources. An urgent shift in regulatory and managerial modes is required to prevent wasteful use of water in this sector and boost its performance (Mellah & Ben Amor, 2016). The conjunction of priority and suboptimal efficiency in the drinking water sector affects water management in the remaining sectors. Moreover, as the size of urban water demand increases over time, then non-strategic crops might face an increasing risk and there may be long-lasting damage to ecosystems and habitats that cannot withstand unstable stream inflow, as ecological allocation seems to be the riskiest one.
Tunisian Water Utility provides water for residential and non-residential uses that include commercial, industrial and tourism sectors. The water authority does not distinguish between residential and non-residential uses with its water allocation system and in practice, non-residential water uses enjoy the highest degree of supply reliability too. Although, it is largely recognized that direct consumption by individuals carries a very high social value that may explain the primacy of individual use purposes among all of the many alternative uses of water, the water authority considers, in addition, that water value in non-residential uses is higher than its value in alternative uses. Nevertheless, non-residential water uses, except tourism use, have been charged at the residential preferential average price that is actually lower than cost price. There is here an obvious deadweight loss that may be remedied with a more appropriate price policy.
The second goal of the water allocation strategy seems to be associated with the food security issue. In practice, this goal is turned into action by ensuring self-sufficiency in some agriculture products. The case of cereals, which are classified in particular as the most strategic crops in Tunisia, attests to this demarche. The government considers cereal a politically sensitive crop, as it is the basic food source of the population that has among the highest per capita wheat consumption worldwide (Ammar et al., 2011). The government is striving for self-sufficiency in cereal production by encouraging farmers to use irrigation and by prioritizing cereal in the water allocation process. Moreover, since 1998, the government has lowered the cereal's water price by half in order to boost production (Tunisian Ministry of Agriculture, 2006). Although cereal production has increased, the cereal import dependency ratio has continued to increase, being at 0.335 in 1990–92, and reaching 0.445 in 2009–11 (Food and Agriculture Organization of the United Nations (FAO), 2016). It seems that production increase does not offset needs growth. The country is fairly dependent upon imports to satisfy its need. Then the cereal water allocation strategy does not seem to be the measure appropriate, ceteris paribus, to realize cereal self-sufficiency goal.
The water allocation in agriculture is not concerned with the market value of water uses, as there is little correlation between the classification of crops as strategic and their value. High-value crops, such as vegetables, have a high degree of water supply insecurity, while wheat benefits from a first-rank priority. There is here an evident source of allocative inefficiency. Additional inefficiencies are generated as a consequence of the lack of initiative to grant priority to users on the basis of their water use efficiency. Moreover, the current allocation arrangement does not create conditions to allow the transfer of water from lower economic value uses to higher economic value uses and does not produce a market-clearing price, since the transfer of water-use rights between farmers is prohibited, which is fostering further inefficiencies.
The tendency toward economic inefficiency under the queuing system of priority, which is consistent with the priority-risk assignment mechanism, has already been highlighted in previous studies conducted in various settings (see e.g., Burness & Quirk, 1979, 1980a, 1980b; Johnson et al., 1981; Howe et al., 1982; Shah & Zilberman, 1992; Zilberman et al., 1994).
Burness & Quirk (1979, 1980a, 1980b) conducted the early theoretical analysis of efficiency implications of the prior appropriation water allocation system focusing on issues of infrastructure, risk aversion, and transaction cost, and they laid out a basic model for efficient model allocation of water. The prior appropriation system, which is a queuing system of priorities developed in the western United States, is consistent with the priority-risk assignment mechanism (Tarlock, 1998). By comparing economic benefits produced under the prior appropriation system with those under an equal-sharing system, Burness and Quirk established that priority leads to inefficiency stemming from the unequal sharing of risk among water users. This inefficiency could be exacerbated by a suboptimal storage policy. The authors demonstrate that in the presence of upstream storage facilities, the water authority tends to store too much water in the reservoir under the priority regime compared with an equal-sharing system although it implements an optimal release policy with respect to the assigned priorities rule. That means that it tends to store water beyond the needs of the senior water user type. Consequently, junior-entitlement holders, who bear the climatic risk, must endure the additional risk related to evaporation and spillage losses from maintaining the dam close to the senior-entitlement holder's capacity. In addition, senior-entitlement holders claim more water when a dam exists than in the case of an uncontrolled river, because their present and future expected profits are guaranteed by the priority rule.
These facts have significant implications on water-use efficiency in agriculture and on the farmers' conservation behavior. Zilberman et al. (1994) report that the water contractors in the western United States, who have lower ranked rights than all prior appropriators, apply less water per acre and their water efficiency is higher than that of the prior appropriators. Burness & Quirk (1979) proved that junior appropriators are more productive at the margin, in the sense that the ratio of marginal revenue to marginal cost increases with decreasing seniority. Thus, junior entitlement holders are more likely to adopt a more productive cultural plan and water saving managerial practices. Senior entitlement holders have the tendency however to under-adopt water-saving technology and then have less incentive to conserve water. Therefore, the reliability of water entitlement seems to have a significant effect on the adoption of water-saving irrigation measures, which is in line with studies on farmers' conservation behavior (e.g., Mushtaq et al., 2009).
In Tunisia, the price charged for irrigation water covers a part of the maintenance and operation costs and does not reflect the real supply cost and the water's scarcity value. The conjunction of prioritization and low water price may result in low efficiency in water use (Chong & Sunding, 2006). In spite of the government water conservation strategy, the technical efficiency of water use remains low in Tunisian irrigated farming and shows room for improvement in the sense that farmers should substantially decrease their water use to produce the same quantity of yields with the same production technology (see e.g., Albouchi et al., 2007; Frija et al., 2009; Chemak et al., 2010; Chebil et al., 2012, 2014; Hanafi et al., 2015). Nevertheless, a further empirical study is needed to bring evidence on the difference in water-use technical efficiency between strategic and non-strategic crops associated with the allocation rule.
Therefore, the current allocation system has not encouraged users to conserve water, as it seems to be possible to considerably reduce water use in irrigated agriculture without altering the level of yields by improving water technical efficiency. Moreover, the rule of priority is justified as a contingent drought management measure, while drought is a recurrent event in Tunisia. Enhancing the capacity of the country to manage drought requires an approach that induces permanent behavioral adjustments to reduce water demands and improve water use efficiency. Therefore, the ability of the country to prevent drought conditions depends on its capacity to improve water technical efficiency, on which the allocation rule has a direct effect and should be designed as an incentive tool for water conservation.
Despite the attractiveness of the reasoning toward food security, a deeper examination of the relationship between agriculture and water policy issues may reveal that the economic value of water in the irrigated production sector is suboptimal. Irrigated agriculture accounts for about 80% of total water consumption. Therefore, a 10% reduction in water use will allow the transfer of this water outside of agriculture, in particular for environmental benefits or would provide the urban sector with a 40% increase in water availability. Meanwhile, if water use reduction in irrigation will cause yields reduction, then this will have a small effect on the national economy and trade balance. Irrigated lands provide one-third of the country's food and one-fifth of exported crops. The export of farm products averaged about 10% of the final sale of Tunisian goods and services during the period 1984–2007 (National Institute of Statistics (INS), 2013). Therefore, a 10% reduction in the irrigated sector's output implies that irrigated land exports will decrease by about 2% that roughly corresponds to 0.2% of the country's total export.
Toward the new institutional arrangement
Since the work of Burness & Quirk (1979), it has been proved that most allocative problems associated with a queuing system of priorities would be eliminated if water rights could be freely transferred. We should note however some points. First, the water market alone could not be considered a fair and acceptable process for allocating or re-allocating water (Syme et al., 1999). Second, market allocation might be suboptimal or inefficient owing to hydrologically related external costs, and legal and social considerations (Brajer & Martin, 1989, 1990). Third, the market equilibrium of water rights will not be affected by their initial distribution if and only if the market operates under ideal conditions. In a more realistic world, the initial allocation of water rights significantly affects the final outcome of the market (Lee, 1999; Weragala, 2010), and an equal sharing of water rights is the necessary condition for Pareto optimality (Burness & Quirk, 1979).
Therefore, the resource's social value of water use right argues for an allocation pattern based on an equitable sharing (i.e., nonpriority) rule but including more market-like incentives. For the agriculture sector, in particular, the water authority could design a proportional sharing system whereby an equal sharing of water allocation reliability is assigned to all uses and under which each farmer is allowed a reasonable amount of water that may be subject to roughly pro-rata cutbacks in time of shortage. Then, water entitlement should be expressed in proportion to stream inflow rather than in terms of water volume, to cope with variability in water availability. Afterward, government permits should be issued to water entitlement holders who are allowed to freely buy and sell their rights.
The nonpriority allocation system might overcome the current system's inefficiency, distribute the risk equally among users and encourage them to adopt water-saving managerial practices. It can be perceived as fairer than the current system as it is capable of producing a more equitable distribution of the opportunities of driving benefits from water, in particular during drought (Tisdell, 2003).
It is worth noting that even if the water price charged to farmers is low, the market price will be their opportunity cost. The market price might create incentives to adopt water conservation technology and to adapt crops plans to meet efficiency requirements (Chong & Sunding, 2006). Moreover, the introduction of a water market should increase the amount of land that a given volume of regional water can serve (Shah & Zilberman, 1992).
The new arrangement proposed is not a complete shift from administrative allocation to market-based allocation but rather a combination of two mechanisms. Counsell (2003) and Rosegrant & Binswanger (1994) record international experiences where the definition and allocation of water permits are achieved by an administrative system with the scope for market-based permit transfers. The institutional arrangements in Mexico, Australia, England and Wales provide interesting examples to inspire the design of water-allocation reform in Tunisia.
Tunisia's water-allocation mechanism is based on ‘priority of use’ assigned by the public water authority in order to secure drinking water for the population, improve self-sufficiency in food and prevent extreme loss in agricultural wealth under drought conditions. The SONEDE, which is a state-owned enterprise, has a monopoly on drinking-water supply across the country. The State guarantees the water resources required to supply the population with drinking water by assigning the first priority to SONEDE demand during the water allocation process.
Regarding the agricultural sector, the water authority currently differentiates between strategic and non-strategic crops. The definition of strategic and non-strategic crops varies according to the agricultural suitability of the region. Without loss of generality, wheat, barley and fodder crops are the most important strategic crops. During a very water-scarce period, however, the water authority tends to preserve permanent crops and livestock. A preferential order is set among water uses for strategic and non-strategic crops. Strategic crop uses are granted the highest rank of priority of all agricultural products, and the vegetable water allocation is usually the lowest ranked, especially during a drought episode.
The Tunisian regulator recognizes environmental water requirements only for the Ichkeul ecosystem and for groundwater. However, these water environmental uses seem to have the lowest rank of priority in the water allocation process.
The development of water transfer facilities has allowed transferring the water of northern regions of the country to coastal regions. Consequently, these facilities have permitted the extension of the application of the queuing system of priority at various territorial scales. In the areas where dams are interconnected and water transfer is made possible, the rank of priority assigned to a particular use is determined by factors arising inside and outside a particular watershed. As a result, the interdependence between regions has been intensified and the competition over water arises between uses type and between regions as well.
The economic concerns related to the queuing systems of priorities, which have been extensively studied in the economic literature, are manifested in the Tunisian water allocation system. The prioritization in water allocation might lead to a suboptimal storage policy and to an unequal sharing of the risk and water-use benefits among users. The analysis also indicates that the current water management and allocation system does not create sufficient incentives for senior water entitlement holders to increase water-use productivity and efficiency.
The Tunisian regulator and the decision-maker can undertake institutional reform to improve water-use performance and social welfare. In the agriculture sector, in particular, a proportional water sharing system that also allows water entitlement holders to exchange their use rights can replace the current queuing system of priorities. This reform may produce a socially optimal allocation and can be seen as an application of the Tunisian water law precept, recognizing that the planning of water use must maximize water value at the national level, it requires, however, the removal of legal obstructions in trading water-use rights.