1

CLIMATE, STREAM FLOW PREDICTION AND WATER MANAGEMENT IN NORTHEAST BRAZIL:

BEYOND MARGINAL GAINS AND TOWARD GAINS FOR THE MARGINAL

Kenneth Broad∗

University of Miami Rosenstiel School of Marine and Atmospheric Science and Center for Ecosystem Science and Policy, and Center for Research on Decision Making, Columbia University

Alexander Pfaff∗

Columbia University School of International and Public Affairs and Earth Institute

Renzo Taddei∗

International Research Institute for Climate Prediction, Columbia University, and Comitas Institute for Anthropological Study

Sankarasubramanian Arumugam

International Research Institute for Climate Prediction, Columbia University

Upmanu Lall

Dept of Earth & Environmental Engineering and International Research Institute for Climate Prediction, Columbia University

Franciso de Assis de Souza Filho

Ceará State Meteorology and Water Resources Foundation

Correspondence Information:

Kenneth Broad RSMAS - MAF 4600 Rickenbacker Causeway Miami, FL 33149 Phone: (305) 421-4851 Fax: (305) 421-4862 Email: kbroad@rsmas.miami.edu

[Submitted to Climatic Change. Please do not circulate.]

∗ Broad, Pfaff and Taddei equally share lead authorship.

2

Abstract. This paper assesses, in aggregate and multi-group terms, the potential benefits from the use of innovative stream-flow forecasts for the semi-arid state of Ceará, Brazil. The many possible applications of such forecasts make this case of general interest, as do the state’s trends in water management policy – from state storage to participatory water allocation to the potential for private water trading. Qualitative analysis of forecast value is informed by extensive fieldwork among policy makers, agriculturalists and other key parties. Its main point is that while specific constraints now limit the effective and equitable application of this innovation, current and potential shifts within the state are likely to increase the forecasts’ value. A better understanding of the conditions under which such forecasts are useful, and for whom, can facilitate public and private choices about generation and dissemination.

1. Introduction

This paper assesses the potential for gains from an advance in climate-based forecasting for Ceará, one of the nine states in Brazil’s semi-arid Northeast. From both aggregate and multi-actor perspectives, we consider the potential for benefits from using one-season-ahead climate-based stream flow forecasts to adjust rates of water release from the reservoirs located in the Jaguaribe Valley, the state’s largest agricultural region. Based upon field observations, of the participatory process in local water allocation and of other private and public decisions about water allocation and use, we identify current constraints on the benefits from forecast use that can be reduced by the policy options currently being discussed in Ceará. Though driven by other agendas, such changes would permit more effective use of the forecast information to minimize negative impacts of water scarcity. We first examine constraints on the aggregate potential gains from forecast use, without regards to who gets the benefits, then consider constraints on gains for different groups in the region, paying particular attention to the relatively marginalized groups.

Ceará’s rural population of 2.1 million, about a third of total population, is mostly employed in agriculture (79%) and is predominantly poor (76%), even by local standards.1 Recurring multi-year droughts have long been identified as a critical factor in the state’s economic, ecological and sociopolitical landscapes (Girão 1986; Prado Júnior, 1989; Parente, 2000, 2002; Neves, 2002; Magalhaes, 2002). A mix of persistent poverty, rudimentary agricultural practices and drought has contributed to ongoing and widespread vulnerability. Vulnerability reduction in the region has emphasized the development of reservoirs, canals and irrigation schemes. These reservoirs are central to rural life but also supply the capital city, Fortaleza, home of 39% of the population. In principle, the impact of improved water management could be broad. Recently, the use of climate forecasts has been offered as a tool to improve the efficiency and reliability of these investments. This motivates our analysis.

The changes in water management since 1987 that are relevant for our analysis have included major institutional shifts, for instance the creation of specialized agencies for different aspects of water resource management and placing the meteorological agency under the auspices of the Secretariat of Water Resources. The decision making structure for local level water allocation has also been changed to include new multi-stakeholder participatory water committees, one for each sub-basin of the Jaguaribe River and for each of the other river basins. These committees were functioning by the mid-1990s.

Concerning this participatory element of water management, each committee makes choices about the annual and seasonal rates of water release from all the reservoirs within the (sub-) basin. Committees choose from a set of scenarios of release rates prepared by the agency for water management, COGERH (Company for Management of Water Resources in Ceará), that operates the reservoir system. Implicit in these scenarios is the fact that major bulk water allocation for the state (as reflected in Figure

3

1) are made by the government prior to the committees’ meetings. In addition, committee decisions are also ultimately subject to a state council. It is these committees, along with all others who have influence over their and any other allocation decisions, who are most likely to use stream flow forecasts.

Much has been written about climate and water issues for this region and recent studies have begun to assess the potential gains from forecast use. Lemos et al. (1999), Finan (2001), and Finan and Nelson (2001) studied rainfed agriculturalists who are not directly affected by the water releases from reservoirs. Drought early warning might be useful for them in planning agricultural and household decisions affected strictly by rainfall, but evidence suggests that the constraints under which these households operate would limit its benefits. Lemos et al. (2002) and Lemos (2003) identified state bureaucracy and politics as a strong influence on the potential value of climate information for any group. Gaiser et al. (2003) present a suite of models of adaptation to climate variation in the region that serves as relevant background for our work but is not focused on forecasts at the seasonal time scale. In this paper, our focus is on the potential for use of climate-based forecasts by users who depend directly upon the reservoir water system, including industrial and municipal water users though we have a particular focus on the irrigated agriculturalists, a relatively poor and vulnerable group. We consider the principal constraints on forecast value for these groups and the role of state policy mechanisms.

As with many technology adoption situations, whether forecasts are incorporated into water management (i.e., reservoir release) decisions is influenced by the details of the decision setting. The involvement of people at the local level in decisions and their implementation has received much attention,2 and it is often claimed that the provision of services will suffer if communities are not closely involved. Within the water sector in Brazil, the push towards participatory water governance has been hailed in terms of democracy (Garjulli, 2001a, b; Garjulli et al., 2002). Yet exactly what form of participation is most effective has not been demonstrated. Cases where the effect of participatory process is shown are relatively few and empirical examination of such participatory or decentralized efforts remains relatively rare (Bardhan, 2002). This paper first identifies specific constraints on forecast value within the current participatory and other processes for water allocation in Ceará and then, in particular, suggests how potential policy changes may affect or shift each of these constraints.

The need to include potential changes in constraints in a forward-looking assessment of the value of a forecast innovation is illustrated by our reconsideration of Sankarasubramanian et al. (2003) for Ceará. The authors make the point that reservoir capacity in the Jaguaribe Valley recently has been increased significantly to address concerns with recurrent droughts as well as interest in high-value irrigated agriculture and plans for new industrial urban centers over the next two decades. Consequently, for the time being the relatively high ratio of storage to demand reduces the value of a season- or year-ahead supply forecasts. We point out, however, that as public and private investments continue in the new major industrial complex and port near Fortaleza and in new urban centers, and as people are drawn out of the vulnerable rainfed agriculture sector into the cities, rising water demand will catch up with current storage investments and the value of forecasts will rise.3 Thus now is an opportune time to think about how water allocation decisions can best be made contingent upon forecast information.

That the value from forecast use depends upon many factors is not a new point (e.g., Hilton, 1981; Glantz, 1982; Changnon and Vonnhame, 1986; Adiku, 1995; Katz and Murphy, 1997; Callahan et al., 1999; Chiew et al. 2000; Pagano et al., 2001; most of these do concern water). However, that point is only rarely addressed through explanations of what specific sociopolitical constraints with relevance for the particular forecast innovation may change. Along these lines, our three foci are: the constraints on water re-allocation not only within the Jaguaribe Valley but also between that agricultural center and the capital city, Fortaleza; individual (i.e., both farmer and water manager) risk aversion; and constraints on addressing equity alongside efficiency.

4

Equity, i.e. the distribution of the net benefits from natural resources along various intra-rural as well as the urban-rural dimension, is a major policy issue. Our analysis distinguishes several groups on the water system (urban, small-scale agriculture, large-scale agriculture, reservoir fishing, aquaculture, etc.) with different wealth levels. We do not focus on the large rainfed subsistence farming communities. They are the focus of previous research and, further, our observations suggest that the changes required for forecast value to rise for these groups are in large part contingent on general socioeconomic development that is beyond our scope to anticipate (for more development-oriented discussions related to the rainfed agriculture group, see Lemos et al., 1999; Nelson and Finan, 2000; Finan and Nelson, 2001; Höynck, 2003).

Our analysis blends disparate research methodologies, particularly from anthropology and hydroclimatology though incorporating economic thinking within the application of each. The identification of potential uses of stream flow forecasts and their implications is done using an optimization model for reservoir release decisions, previously developed (Sankarasubramanian et al., 2003), that we summarize in Section 3. That work provides the claim that measurable gains in water availability can result from changing water releases if stream flow forecasts are well used. However, as Sankarasubramanian and Lall (2004) emphasized, it is not designed to reflect how various societal groups vary in their influence upon the use of forecast information and in how they are impacted by water management. Thus our field methodologies are intended to examine possible gains from forecast use in a broader social context. The fieldwork also suggests how best to identify parameters needed for more informed reservoir optimization using forecasts.

Our conclusions rely most heavily upon: archival research; participant and extensive field observations from 2002 to 2005; structured and semi-structured interviews; and database analysis of over 4,000 records concerning land plots within the Jaguaribe Valley. Taddei et al. (2005) provide a more complete description of the relevant stakeholder groups. This research has been done in collaboration with members of relevant Ceará agencies, water user groups and civil society organizations involved in water allocation.

We are not advocating or discouraging any of the potential changes that we identify. There will be many reasons why they will or will not occur, and equally many views on whether they should or should not occur. In any case, those interested in evaluating the potential utility of the climate-based forecasts that we consider should be aware of the potential for changes that could affect forecast value. We stress that in thinking about value, it is important to consider not only aggregate total value but also the distribution of impacts from any change including forecast use.

Below, we first describe the history of water management (Section 2) and then the reservoir optimization advance being considered for operational use in Ceará (Section 3). We then present analyses of constraints and their evolution (Section 4) and finally our conclusions (Section 5).

2. Water Management in Ceará

2.1. DROUGHT, DEVELOPMENT & WATER MANAGEMENT HISTORY OF THE STATE

The state of Ceará, in Brazil’s relatively highly populated4 and relatively undeveloped semi-arid Northeast region has over 7.4 million inhabitants5 [Fundação Instituto de Pesquisa e Informação do Ceará (IPLANCE), 2002b]. The fraction of gross domestic product (GDP) from agriculture has dropped from 30% in the 1950s to around 7% (industry and services in Fortaleza are responsible for 85%). The rural illiteracy rate decreased to 56% in 1999 (Governo do Estado do Ceará, 2000b). Small

5

holdings of less than 10 hectares represent 70% of holdings but only 5.4% of area. State GDP has grown in the last 15 years but, for most, agricultural productivity has been low.

Low productivity in agriculture has been attributed to periodic severe droughts, poor soils, skewed land distribution, low levels of education, high levels of poverty and underemployment, and limited physical and social infrastructure (Costa et al., 1997 p. 138). Climatic variability has permeated life, with links to fundamental issues such as: cyclical migration (Montenegro, 2001; Neves, 2002); religion (Villa, 2000; Della Cava, 1970; Couper-Johnston, 2002); economic and demographic instability (Della Cava, 1970; Parente, 2000; Greenfield, 2001); the relations of the State with its poorer inhabitants (Neves, 1995, 1998, 2003); and even the ties among elites in public office (Faoro, 1984; Parente, 2000) who have gained from “the drought industry” linked to federal drought assistance (Callado, 1960; Cunniff, 1975; Coelho, 1985; Medeiros Filho, 1988; Kenny, 2002; Albuquerque Júnior, 2004). These agricultural constraints and also these longstanding relationships clearly affect the incentives to respond to crisis in the region through seasonal or longer-term changes in investments. They also affect uses of information.

Any brief chronology must include at least two of the extreme droughts that shaped perception and action. The drought of 1777 is said to have caused the loss of almost all of the State’s cattle, ending a short economic boom in which Ceará had become the most important beef producer within this Portuguese colony. The severe drought between 1877 and 1879 is alleged to have killed over 500,000 people (Neves, 2000; Greenfield, 1986, 1992, 2001; Davis, 2001).6

This latter Great Drought of 1877 transformed drought from a private to a public matter (Villa, 2000). The imperial (later federal7) government decided to invest its best technicians in the fight against drought using science. Migration to other parts of the country was induced to a limited extent, but the focus was water storage.8 The construction of massive reservoirs began under Emperor Pedro II in 1886 and was continued by the federal Inspetoria de Obras Contra a Seca (IOCS or Inspectorate of Works Against the Drought) in 1909.9 IOCS later became Departamento Nacional de Obras Contras as Secas (DNOCS), whose responsibilities are mainly research on infrastructure and operations of the reservoirs and whose history is meshed with that of Ceará. In particular, its past actions have been linked to the state’s political clientilistic relationship with the elite and poorer classes. It has been claimed, for instance, that infrastructural investments have been made to benefit local elite groups (Coelho, 1985; Villa 2000; Albuquerque Junior, 2004).

The rural oligarchy essentially maintained control in various ways until 1986, when they lost the state elections to a new group of young businessmen focused on industry. Development focus gained prevalence, and incentives were created for economic progress. In 1987, the State Secretariat for Water Resources [Secretaria de Recursos Hídricos do Ceará (SRH)] was created and the Fundação Cearense de Meteorologia e Recursos Hídricos (FUNCEME, i.e., the state meteorology and environmental research agency created in 1972) and COGERH,10 a new state water agency set up as a mixed-capital company, were placed within it. The latter was to help make up for a financially weakened DNOCS. COGERH’s responsibilities include oversight of the debate over reservoir releases within the new participatory water management process being created within the same reform.

In 1992, state law 11.996 created a state system for management of water resources. It called for water planning and management to be integrated, decentralized, and participative. Management was to include the licensing of and charging for water and permits for infrastructure development, though there has been minimal enforcement of these regulations up until this point in time. Regarding participation, at least a formal partial decentralization of water management was effected. Conselho Estadual de Recursos Hídricos (CONERH) or the State Council for Water Resources was given the ability to arbitrate water conflicts, although oversight remained with the state’s judiciary, and a mandate was given

6

for one water committee for each of the eleven river basins (or sub-basins in the case of the Jaguaribe, due to its length). These were intended to have the power to decide on the allocation of water inside of their (sub-) basins, though under the power of the state council.

In 1994, just as the state was to experience a crisis in water supply, a participatory water allocation group was created in the Jaguaribe Valley. This group would eventually evolve into codified watershed committees. The 1994 group was to decide how to allocate water in order to reduce political conflict due to any impending water shortage. In 1995, the first watershed commission was formed in a small basin called Curú. Seven of the eleven river basins now have water committees that decide on water allocation. Usually a committee allocates the water of the largest and most important reservoirs of the basin, a potentially highly contested choice. Diverse watershed stakeholders meet each January and June and choose from scenarios (generated by COGERH) showing the rate of decrease in reservoir water storage for each of a set of rates of released stream flow for the upcoming six months from all of the largest reservoirs in the watershed. In small reservoirs, local commissions were also created, with some effort to mimic the composition of these basin committees: 30% of members from local users; 30% civil society; 20% municipal government; and 20% state and federal governments.

National actors have weighed in again recently. The national water agency, Agencia National das Aguas (ANA), was created in 2000 (Federal Law 9984/2000) to implement modern water management focused on transboundary watersheds.11 National impacts on Northeast Brazil may in part come through turnover within the political cycle. The heads of even the most technical agencies are often replaced after state elections and agencies’ highest offices are filled with an eye to party lines. After the 2002 elections, for instance, key changes occurred at national and state level,12 with implications for the organization of the state water system. The responsibility for organizing the water committees was firmly centralized in the SRH, while revenue generation to cover water system costs became a priority.

2.2 THE JAGUARIBE-METROPOLITANA REGION

Our research focus is the Jaguaribe-Metropolitana (JM) area.13 The Jaguaribe River system (Figure 2) is the source of water for over forty-five municipalities. All of the most important economic centers of the Jaguaribe Valley are among them. The valley occupies about half of the area of the state and is home to just over half of the state’s hinterland, or rural, population. Its occupants range from rainfed agriculturalists to large agribusiness enterprises to urban dwellers.

The JM hydrological system is dominated by reservoirs and a series of flood gates that are operated by COGERH (Figure 3). Water demands include: human consumption in Fortaleza and in the small towns in the hinterland; growing agribusiness; small family farmers; shrimp aquaculture (which has large environmental impacts of growing concern14); marginalized riverbed farmers using sump pumps; and poor fishermen who require specific reservoir levels to be productive.

Political alliances exist at the sub-basin level, e.g. among the communities associated with a particular reservoir, and organization of water stakeholders has been in large part by location. For instance, communities located upstream or downstream of the reservoir gates may disagree, as the former tend to oppose water release but the latter tend to favor them. Organization also occurs by occupation, e.g. unions, cooperatives and associations.

The water allocation commission of the Jaguaribe River has 107 members, from four of its sub-basins (Alto Jaguaribe, Médio Jaguaribe, Baixo Jaguaribe, and Banabuiú15). Of the 30% from civil society, i.e. various associations of inhabitants, 2/5 are from the rural workers unions. As noted above,

7

only a clear minority (here 28%) are the local water users, e.g. companies providing water for human use, producers from the public and private irrigation areas, and other associations. The other 42% are from the municipal governments (25%) and 17% from the state and federal agencies, i.e. DNOCS, COGERH, Empresa de Assistência Técnica e Extensão Rural do Ceará , Secretaria de Agricultura Irrigada do Ceará (SEAGRI), and Companhia Hidrelétrica do São Francisco. Looking across space, the Baixo Jaguaribe sub-basin has the most representatives (36%). It has less rice but more fruit and shrimp and also bigger and more organized producers using more sophisticated irrigation (Figure 4).

Again, users are only a quarter of the commission’s members. Adding rural workers’ unions, who represent small farmers working on others’ lands, as well as farmers who cultivate reservoir and riverbed lands, brings the total up only to 40%. Two stories are commonly offered to explain this structure. Initially, COGERH believed that all of civil society are indirect water users and direct producers of water pollution. Alternatively, and as suggested by the relatively large number of representatives of local political elites (i.e., local governments) and state and federal agencies, COGERH wanted legitimacy among local level politicians and feared (as did top government officials) a loss of control of the water allocation process to non-governmental stakeholders (Taddei, 2005).

3. Climate-Based Stream Flow Forecasts in Water Management

The development of Fortaleza’s industrial and service sectors depends on the provision of water from around the state, with the Jaguaribe River being most important. As there are no naturally perennial rivers in the state, the official and private investments in infrastructure throughout the 20th century were “to make water” (in the local phraseology) by accumulating large amounts of water during the rainy season (January to June) for use in irrigation during July to December.

Given these investments, which themselves have reduced vulnerability, a recent idea is that the use of forecasts could help to better manage this water storage system.16 Research revealed two primary factors that control regional precipitation, the state of the tropical Pacific [El Niño-Southern Oscillation (ENSO)] and the tropical Atlantic sea surface temperature. Extreme droughts are often but not always associated with the extreme El Niño events (ENSO warm phase). Whether they are or not depends partly on the state of the Atlantic and on other stochastic weather factors.

There is now some ability to predict general patterns of climate variability, which for this case means the ability to predict stream flows. Such a sense prior to the dry season (i.e., July to December) of how much water will flow into reservoirs in the next rainy season (January to June) should, in principle, permit adjustments to reservoir releases that provide net benefits. This information should help to make tradeoffs about reservoir operation that take into consideration both the short-term, i.e. immediate, and longer term, i.e. next year’s, demands for water (see Figure 5).

Ceará’s current reservoir operating policies can be considered conservative, in the sense that they currently assume zero inflows into reservoirs from future rainfall. The policies specify a volume of water to keep in reservoirs at a given date and a set of prioritized releases to specific users. The latter include demands for drinking in Fortaleza, if Jaguaribe water is necessary to meet those needs (see Figure 1 for bulk water allocation across the state and Figure 6 for a breakdown of water users within the Jaguaribe Valley).

Assumptions about inflows to reservoirs generally are based upon historical records and can be used, with current storage, to assess likely water supply for the coming season. Recently, it has been learned that remote climatic conditions such as tropical sea-surface-temperature anomalies can be linked

8

to hydroclimatology to yield predicted stream flows (i.e., the inputs to reservoirs) at seasonal and longer lead times, (Hamlet and Lettenmaier, 1999; Sharma, 2000a; Piechota et al., 2001; De Souza and Lall, 2003). This could benefit water management (Dracup and Kahya, 1994; Cayan et al., 1999; Sharma, 2000b; Goddard et al., 2001). However, only a few studies have investigated the potential benefits that could be obtained in water systems management by utilizing such forecasts of streamflow (Yao and Georgakakos, 2001; Hamlet et al., 2002; Arumugam et al., 2003). For Ceará, the question is whether the gains from using estimates of future inflows are sufficient to drop the conservative zero-inflow assumption.

To address this question for Ceará, Arumugam et al. (2003) use a model of the state water system to analyze the potential for gains in reservoir performance from the use of streamflow (i.e. reservoir input or inflow) forecasts. Forecasts are obtained using exogenous climatic indices that denote the slowly varying anomalies in sea-surface temperatures over different parts of the globe that influence the moisture transport into a region on a seasonal or long-lead basis. Such forecasts based on climatic conditions outside Ceará can be obtained using either dynamical or statistical modeling approaches. The forecasts used here are statistical, following the forecasting methodology in De Souza and Lall (2003).17

To study the Orós reservoir, a multivariate, semi-parametric conditional resampling strategy was employed for forecasting annual and monthly inflows into six reservoirs in Ceará using the Nino 3.4 index (an indicator of El Niño and Southern Oscillation) and a North Atlantic Dipole index. De Souza and Lall’s (2003) streamflow forecasts correlate well with observed flows into those reservoirs. A forecast is issued in July for the ensuing 18 months, to December of the next year. July through December in the current and next year typically have zero flow, so only the monthly flows from January through June of the next year are forecast. The reservoir system is simulated from July through June, to determine the potential to meet the annual demands for irrigation (which occur predominantly during July to December) and for municipal and industrial supply (approximately constant for each month between July and June). The reservoir-allocation model is similar to those in Lall and Miller (1988) and Lall et al. (1996), modified with an end-of-year storage constraint as well as an annually varying yield at a specified target reliability for each use category. This yield or contracted release in principle could be promised to specific users. Fees and compensation for non-delivery of water could depend on the reliability assigned to the contracted release.

Sankarasubramanian et al. (2003) compare the use of four different inflow-forecasting approaches for the period 1950 to 1996. The first is the De Souza and Lall (2003) approach. The second is a “climatology” based forecast, where historical years from the period 1950 to 1996 are drawn with replacement, with equal probability to form an ensemble (assuming no skill in forecasting). The third method is a forecast of zero inflow for the ensuing year at all reservoir locations, i.e. the method currently used.18 In fact, while inflows near zero were experienced in about 1% of the years in the historical period, the probability distribution of inflows is strongly skewed towards small annual inflows, with a few dramatically larger flows. The fourth forecast method is a perfect forecast, i.e. the observed sequence for the subsequent year is used.

Given the streamflow forecasts, the model allows one to set the end-of-year target storage for each reservoir, to meet the demands for the next year with a specified reliability, and to set both minimum and maximum releases by use and reliability level. The water-allocation optimization model then generates yields for each of several potential water delivery contracts at specified reliabilities that maximizes a weighted combination of the releases for all contracts. The weights are presumed to reflect the utility associated with each type of use (e.g. Municipal or Irrigation) at the associated reliability level. In addition to the maximization of some economic value (Hamlet et al., 2002; Yao and Georgakakos, 2001; Faber and Stedinger, 2001), Arumugam et al. (2003) also assess performance in terms of reservoir ‘reliability’, ‘resilience’, and ‘vulnerability’ shortfalls relative to demand (Hashimoto et al., 1982), and spill

9

events for a range of years. Simulations for a fixed reservoir size illustrate that as forecast skill increases, yield rises at any level of reliability, while reliability rises at any level of yield. The K-nearest neighbor forecasts used for Cearà have considerable skill, with a forecast-to-observed-flow correlation above 0.7 under cross validation, and operational performance using these actual forecasts is consistent with the simulation results for this level of skill. This improvement in yield (for any reliability level) using existing 'K-nn' forecasts is achieved without a reduction in the other performance measures considered, resilience and vulnerability.

4. Shifting Constraints and the Value of Climate-Based Forecasts

The forecast innovation described above moves beyond just generating climate-based forecasts to modeling their potential use by the operator of a reservoir and the aggregate effects of such forecast use on societal outcomes. Such modeling efforts can quantify the value of forecasts once the objective--who matters across a few broad groups and what do they care about--is specified.

Such modeling also shows explicitly that multiple factors influence the value of forecast use for any given objective. For instance, using an objective that distinguishes three sectors – industrial, agricultural and human consumption - Sankarasubramanian et al. (2003) find that recent increases in reservoir storage limit the value of stream flow forecasts. Yet we argued above that in the future these forecasts will have significant value as development proceeds and water demands increase. That is, we identified a likely shift in an existing constraint on forecast value (‘surplus’ storage) that would raise future forecast value.

The sub-sections below list other factors that constrain the net benefits from use of climate-based forecasts of stream flows into reservoirs. The first sub-section considers aggregate gains from forecast use without regard to who gets benefits. The second sub-section considers the perspectives of particular groups on the use of forecasts, since there are many distinct groups whose objectives may conflict regarding water allocation. In each sub-section and for each factor listed we describe the current situation, with constraints, and then based on our field observations suggest what changes may permit forecast benefits to rise.

We emphasize that we neither advocate nor argue against any particular change we identify. Some of the potential changes concern highly contested issues involving many actors and views including about distribution and fairness. Our own analysis stresses the need for evaluations to consider not only aggregate outcomes but also the distribution of impacts (see more below). In any case, we are simply commenting on the implications for forecast value of possible changes.

4.1 BEYOND MARGINAL GAINS

4.1.1. Water Allocation Mechanisms in the Jaguaribe Valley

One gain from the use of forecasts is more water for society to use. For instance, benefits can include lower reservoir overflow and lower evaporation if, in anticipation of rain, withdrawals can be planned and water allocated. The value of forecasts thus is linked to the value of water.

Water’s value in turn is affected by how it is allocated. If water is allocated to the highest-valued uses, then having more water will add the most possible value. We are using ‘most’ and ‘highest’ here in

10

an aggregate sense, as typified in the calculations of the state’s GDP or its total tax revenues or foreign exchange receipts from exports that are of interest to many local actors.

A commonly used mechanism for allocating a resource to its highest-valued uses is trading among potential users. Currently, water allocation in the Jaguaribe Valley does not feature such trading and water does not go exclusively to high-valued uses. Formal water rights are virtually non-existent and any water trading after the official water allocation is limited and informal. Further, misgivings about formal water trading have been publicly expressed (for an example, see Figure 7) by actors who are concerned with the potential inequities in the distribution of water.19

However, the situation may change. A policy experiment in 2001, Aguas do Vale (Waters of the Valley), paid rice growers approximately $25/hectare to not produce.20 Water savings went to perennial cultures and shrimp production. This was done in light of concerns about water conflict in a year of low rainfall and appeared to head off conflict. It achieved a temporary re-allocation of water, though it did not achieve a lasting shift out of rice in the areas that received payments.21

Difficulties existed in collecting payment from those who received water (only 21% paid). This presages challenges for new water institutions that link water transfers to payment.22 On the other hand, that this experiment occurred at all suggests significant institutional interest in this kind of behavior and is seen as communicating that such water trading may become the norm.

The national water agency (ANA) is the actor most often cited for support of water licenses (outorgas) that are natural precursors to trading as well as of tariffs, although there has also been activity in the state and the national legislatures regarding the reform of water laws to allow for licenses and billing.23 Since it is not easy to shift the regime for allocation of such a fundamental resource, national signals matter. Water laws recently enacted in Ceará allow for transference of outorgas among users, but there is very little public discussion of this aspect of the legislation and our observations indicate that the public is generally unaware of this (it may not be stressed by lawmakers given the fierce political debate concerning any link to a free market system).24

Signals to the public in Ceará since the 2002 election have also been in the direction of payment for water. Rationales given include incentives for allocation to higher-valued uses and “cost recovery” (e.g., balancing COGERH’s budget). In any case, Ceará’s SRH has taken the idea seriously enough to conduct studies through the Ceará Federal University (UFC) of the ability of the farmers within the Jaguaribe Valley to pay for water. SRH also promotes the possession of water licenses, for instance as preconditions for Banco do Nordeste do Brasil (Northeast Brazilian Bank) loans. If having an outorga becomes necessary to use water, Ceará will be close to water rights or claims that can be traded. Price could guide such reallocations, at least among industries, municipal water and sewage companies, and agribusiness enterprises. For equity, as discussed in 4.2, the state may wish to protect small-scale agricultural producers. It is possible that providing some type of license,25 for water to use or to trade, would achieve this.

Trading could well occur without the considerable institutional complications from billing and payment. If water rights--who controls the use of the water--are understood, then water trading may well arise on its own. There are already informal examples of this within Ceará.

A regional decree from the Cariri area in southern Ceará (and more specifically within the Batateiras Valley) that became effective from 1870-1880 gave control of water to owners of the lands containing the water sources. Since then, inhabitants have been selling their water to each other and to local municipalities (Kemper et al., 1999; Campos and Simpson, 2001; Garjulli et al., 2002). Despite the fact that this practice violates state and federal legislation, there appears to be no official enforcement of

11

those laws. To this point, water claims are defended through social pressure linked to traditional practices and sometimes force. Given de facto rights, agricultural producers created on their own a system of trading that by now is more than a century old.

In addition, within a publicly managed irrigation area inside the Jaguaribe-Apodi perimeter, during water scarcity growers of perennial cultures (mostly fruits) negotiate with beans, soy and corn growers for the water implicitly allocated by the local water committee (and by the internal water distribution schedule of the irrigation area). A form of informal water right seems to exist.

The latter example is worthy of note concerning distribution. While trading may ‘be stacked’ against the more poor, as some advocates suggest, if the water allocation committees do create effective water use rights then the ability to sell water may lower the poor’s vulnerability during droughts. Surrendering their water clearly takes away their ability to produce, but appropriate compensation and the time to generate income in other ways could make this result preferable. This point also holds in principle at a larger scale, as discussed just below. To some extent, such water rights could also incorporate the poor in addressing risk, as discussed further in 4.1.3.

In sum, water transfers to higher valued uses will raise forecast value. Further, movements towards transfers are present in Ceará. Finally, they could be done taking equity as a constraint.

4.1.2. Scope of Participatory Water Allocation Decisions

This section also concerns transfers, water value and thus the value of water-saving forecasts but for the participatory (versus private or market) setting of Water Allocation Committees (WACs). The value of providing forecasts to inform WACs’ allocation choices depends on the potential value from the water re-allocations they are considering. Thus the scope, e.g. the spatial scale, of the WACs’ or others’ choices for which forecasts are provided will constrain the value of those forecasts. We believe the trend is toward a larger scope for such choices and thus higher value.

The WACs’ history helps to understand the current scope of their participatory water process. Their discretionary scope is limited, as seen in the small differences among the release scenarios generated by COGERH for WACs to consider (See figure 8 for COGERH scenarios).26 This does not mean that in principle the WACs or analogous groups could not have a greater say in water allocation. Yet the current process constrains the value of any forecasts provided to them.

Many discuss WACs (in Ceará and elsewhere in Brazil) in terms of democratic decentralized water decision making, but their limited scope may be better understood if it is recognized that in Ceará their existence appears to be due in large part to a desire to dispel and to manage potential rural anger about water re-allocation. The Ceará WACs were created shortly after significant discontent concerning water transfers to the capital city of Fortaleza during a 1993 drought, especially in the Jaguaribe Valley. Many agricultural water uses were prohibited within the Valley during that water crisis, a ban that was at times enforced by the army (Garjulli et al., 2002). Meeting the demands for drinking and other basic urban water needs in this fashion was, not surprisingly, very unpopular in the Valley. Thus the state felt that something had to be done.

Describing a WAC in this way, i.e. as an instrument for management of discontent, is at odds with the point that any popular institution breaks from longstanding paternalistic relationships between the state and marginalized rural inhabitants. Yet despite this fact, there does not appear to have been a clear break from “business as usual” in water. That is most starkly manifested in three ways: first, the

12

decisions of the WACs can be overturned by CONERH, a water council controlled by the government; second, WACs include many non-water-using actors including actors from government; and third, the WACs must choose from the scenarios generated by COGERH.

Evidence from our interviews with actual and potential WAC participants and beneficiaries supports the thinking that not only do elites retain significant control over water allocation but also those in the Jaguaribe Valley understand that fact. We have observed significant mistrust and ambivalence about the participatory allocation process, though some observers and some participants view the process overall as a leap forward in terms of democracy (Garjulli et al., 2002; Taddei, 2005). Over the span of our project, trust in the process appears to be growing.

Yet we emphasize COGERH’s control over the scenarios that the WACs consider. While many COGERH employees work passionately to bring interested parties to the WAC meetings and, generally, to broaden popular participation,27 it is still the case that the water considered necessary for Fortaleza is removed from consideration prior to WAC choice. This significant limitation on the scope of WACs’ choices has strong implications for not only how effectively democratic water allocation is but also the value of the forecasts. To have greater value, the forecasts would have to influence the larger-scale decision about urban versus agricultural use of water, either within a ‘WAC with greater scope’ or within an agency taking such decisions.28

Such an ‘expanded mandate WAC’, or other institution making broad allocation choices in a formal, public fashion, could actually come into being relatively soon in Ceará. It could benefit from forecasts. Such a group would draw members from the valley and the city (both drinking and industrial users).29 A major event that could easily precipitate its formation is the ongoing construction of the Canal de Integraçao (“Integration Canal”) from the Jaguaribe Valley to Fortaleza and the new port in Pecem. Its completion will make feasible a much larger scale of water transfer, necessitating a focus on a new set of water transfer choices. Just as for the choice made currently by WACs, for these new transfer choices about how much water to move to Fortaleza from the Valley the best existing forecasts about stream flow in the Valley should matter. At this scale, including for high-valued uses, potential forecast value would be higher.

To whom the benefits of forecast use would go is again worth a note. In an expanded WAC participatory setting, presumably compensation for water transferred out of the Valley to the city would be negotiated. For a vulnerable region, just as for any vulnerable household, surrendering water in exchange for compensation could be preferred, at least when forecasts suggest drought.

Yet forecasts could also be used in decisions made beyond these WACs’ reach.30 Some of the politicians and agencies responsible for economic development may favor guaranteed water for activities with fiscal benefits, influenced by pre-existing competition among states to attract investors (see Tendler, 2000 for discussion of the “Fiscal Wars”).31 If this leads to new major transfers along the new canal it could precipitate agitation (as in the 1993 crisis) and a public negotiation. However if some form of individual or regional water rights in the Valley were to be created formally or at least acknowledged functionally, reallocation might automatically require compensation to those surrendering the water. Thus the gains in welfare from reallocations – be they due to forecasts, economics or politics - could be distributed. In sum, an expanded scale of water transfer decisions will raise the value of forecasts provided to the participatory committees. Again, any institutional innovations required could be designed to take equity as a constraint.

13

4.1.3. Risk Perception and Forecast Non-Use

The non-use of forecasts would of course eliminate the potential gain from their use. Further, there is good reason to think that risk aversion leads people to prefer not using forecasts in making decisions, and risk aversion itself is not likely to vanish. However, innovations in institutions that can reduce both perceived and effective risk could affect such decisions.

At the household level, we have observed ‘defensive’ or ‘hedging’ behavior in the decisions by the more vulnerable populations to invest, or not, in fruits or advanced irrigation equipment.32 Such behavior may follow an individual preference for a lower variance in agricultural outputs. Those who lack resources may neither easily survive a weather-related fall in their output nor be capable of taking full advantage of a good weather shock to increase production and thus prefer a safer, even if lower, average level of output. Further, even if the forecasts have improved and thus more people could in theory gain from their use, hedging within seed and crop choice and occupational multiplicity (Comitas, 1973) may now be deeply ingrained habits.33

The state’s release regime from the reservoirs also suggests that risk aversion is paramount. Currently, the releases for the dry season are determined under an assumption of zero rain in the upcoming rainy season. That is sometimes observed but is not the long-term average – it is observed in less than 1% of years in historical records. The state’s use of this pessimistic estimate suggests an asymmetry in effective societal preferences, i.e. lower weights on the ‘upsides’ versus the ‘downsides’ of outcomes.34 This could simply reflect the risk aversion discussed above of the household constituents served by reservoir managers. It could also reflect a perception by water managers that they will not be rewarded for having a bit more water for production but will be punished if they come up short relative to basic water needs. Political benefits and costs faced by their bosses could be even more asymmetric than households’ views.

Water insurance could help address concerns with risk, in this as in many other settings.35 For insurance to be financially self-sufficient there are charges for water, i.e. contracts between the water agency and users. As noted, this is already common in Fortaleza and under consideration in agricultural areas. The insurance twist is that some form of temporary ‘right’ to the water is conveyed by a delivery contract: if water promised is not delivered, compensation is paid. This could create value in two ways. First, being able to count on compensation when production of output has been undermined by drought conditions could facilitate investments by farmers that would raise their average output. Second, the value that we have suggested forecasts can create would be more likely to be realized, since the effective risk from their use would be reduced.36

As noted, many are concerned with the distributional implications of any system that charges for water, potentially even one that explicitly provides compensation for water non-delivery. In light of these concerns, two types of precedent are noteworthy: first, public service provision has often subsidized the participation of the poorest, around the globe; and second, Ceará facilitates participation in education via payments per child. Thus, any actual water insurance program that is implemented in Ceará could well be designed to include low or zero charges for the poorest.

Given the centrality of uncertain water supply in this region, it is not surprising that not only the preconditions for insurance (water billing and collection) but also insurance itself are being discussed among international development groups and Cearense academic and government research groups. However, government may resist policies that involve promises of water delivery. Yet, while the situations differ, it is worth noting that the government has long been willing to provide rainfed harvest insurance. Though the amounts that are provided in bad times to the most vulnerable are extremely low, nonetheless this policy enjoys widespread popularity.

14

Even without insurance, effective perceptions and impacts of risk would change if the water trading discussed above does occur. It is commonly believed that risk aversion decreases with wealth. The wealthy can better weather the downsides of risk and move quickly to benefit from the upsides. As a result, often they lead in adopting innovations such as more advanced and costly irrigation technologies. In the Jaguaribe Valley, irrigation methods do differ by wealth.

Of relevance for this study is that producers of higher-valued crops are often the relatively wealthy and as discussed above they are more likely to get water if water trading occurs (though, as noted above, such trading could lower vulnerability of those who sell their water). This holds not only within the Jaguaribe Valley but also when water is transferred from rural to urban users including corporations. Thus, water trading to higher-valued uses can effectively lower aversion to risk in average water use, as more decisions are made by the relatively wealthy. This makes the use of forecasts in water management more likely, increasing the value of the forecasting.

Finally, a novel combination of forecasts and trading could help those who do not wish to face the risk of uncertain rainfall, such as the more vulnerable (this point follows Osgood et al., 2005; and a related policy suggestion in Brown and Lall, 2005). When stream flow forecasts are issued, i.e. before rain occurs, the trading of claims on the water (or ‘places in a water line’) can in principle provide a useful early signal of water scarcity that helps people to make decisions. 37

Those who have high-value and vulnerable crops such as fruits could buy the highest-priority claims, at the front of the line, in order to almost surely get water even if rains turn out to be low. This reduces vulnerability and they are likely to pay for that. For those considering investments in fruit crops, the price of such high-certainty water is a signal that comes before early decisions for the season. This can coordinate decisions. If many are considering the same investments and bid for high-certainty water, its price will be high. This can help to avoid situations in which more people plant such vulnerable crops than there is water to support and somebody loses.38

Early signals have value even for those not planting fruits. Many farmers make early choices about crop, seed, method and area. Some consider alternative urban employment. Early trading of water claims can coordinate farmers’ plans, conveying the water implications of their and fellow farmers’ plans that affect demand for water quantity and certainty. The price of a position in the water line signals how much all others value certainty, or fear uncertainty, and thus signal the competition for the rain when it comes. This informs choices that determine vulnerability. For instance, when the forecast is of low rain and the price of a water claim is sufficiently high, should he possess an initial water right or claim and understand that uncertainty about the rain is high for the coming season, a vulnerable farmer could act on this early signal (which combines demand for water information with the forecast of supply) to decide to work in the city that year.

If the vulnerable do not have pre-assigned rights that they could sell, then the government could reinvest some of the revenue garnered from selling high priority rights in support of the necessities of smallholder agriculture (e.g., in the US, governments weigh in financially to assure water for environmental outcomes like fish survival, which do not have bidders in the trading). They could also transfer the revenue as a form of general compensation outside of the formal trading, e.g. through technical help with crop switching or incentives for sectoral switching.

This sort of innovation certainly has not yet been discussed widely in Ceará. Yet the large producers of fruits are more likely to have water than are producers of rice for subsistence, as evident in historical experiences like the Águas do Vale program. Further, they seem to know that this will be the case. This suggests that the potential gains from trading and signaling of this type are apparent. Their formalization is not out of the question, following the current active consideration of the use of stream flow forecasts

15

and discussions of both water trading and insurance. More generally, it is clear that very competent agencies in Ceará are considering the application of analytic innovations for practical benefit, i.e. ‘leaping forward’ in management.39 Overall, all of the innovations discussed above would raise the value of stream flow forecasts.

4.2. TOWARD GAINS FOR THE MARGINAL

Our focus above was the potential for aggregate gains, for understanding long-term regional development opportunities. This left aside how forecasts’ net benefits are distributed among the stakeholders in the Jaguaribe Valley. Competition for water among upstream and downstream groups is endemic, for instance, and it would surely affect perspectives on the use of forecasts. Such differential impacts of release decisions imply that even innovations with aggregate gains could hurt some users. This almost surely will lead groups to contest the use of such forecasts.

Even if a forecasting innovation brings small changes, any relatively marginalized group may still benefit or lose enough for them to care a great deal. In this valley, fisherpersons, reservoir and riverbed farmers, and others are directly impacted by the WACs water-allocation decisions. If stream flow forecasts are employed, what impacts will that have for all of them? In addition, can either stream flow or precipitation forecasts potentially bring gains to rainfed agriculture?

4.2.1. The Poorest Have Only Rain

With unchanged infrastructure, stream flow forecasting and reservoir releases will not directly affect the rainfed subsistence farmers. Those 800,000 to one million subsistence farmers (versus approximately 50,000 people living from the irrigated lands) could in principle benefit if they migrate to work on lands that are irrigated. 40 However, their own agricultural outcomes are not affected by irrigation, so the types of innovations discussed here will not be of much use to them.

Things certainly should change over the next few decades in Ceará. State development plans, as noted, call for six new urban centers with employment opportunities intended to attract the rainfed population and thus reduce their vulnerability. While older farmers were not educated, investments in education have yielded very high literacy among the young, facilitating a shift. However, predicting the pace of development of the state is beyond our scope. Consequently, we do not make statements about how eventually forecasts may affect the rainfed agricultural group.

4.2.2. Distribution Across Irrigated Groups

Vulnerable producers who depend on the water storage and irrigation system may care a great deal about even the relatively small amounts of water that the WACs have discretion to allocate. For instance, the output (and consumption) of those who fish in the reservoir is very sensitive to reservoir levels. The same is true of those who do their farming at the edges of reservoir waters.

In addition, riverside agriculturists are limited by their pump technology to work within certain distances of the waterline, such that they too are affected if waterlines shift either up or down.

16

These points are consistent with our observations of WAC meetings. During periods of significant drought, attendance by the vulnerable rises. During the immense rains of early 2004 the reservoir farmers, essentially alone among all groups, remained highly concerned about water release rates. The reason was that low release rates, and consequent full reservoirs, left their desired lands under water. Their concerns dominated recent WAC meetings (Taddei, 2005).

These producers certainly carry little weight in total output of the state or even of the valley. However, their political weight may be greater, just as it is said that the invasion of grain stores by rainfed subsistence farmers during tough times is a highly public signal that gets responses.41 We have observed that efforts to press the agendas of these groups goes beyond WAC meetings. One dramatic example is that some groups have broken the locks on reservoirs in order to effects releases themselves. That occurred most recently in July 2004, when farmers took control of the Lima Campos reservoir.42

The simplest point here is that WAC or other allocation decisions should take the outcomes of these groups into account in some way, i.e. they wish to have a place within the objective.43 While the effective weight they receive in society will be known only after negotiations occur, within any formal modeling of the gains from forecast use, these groups can be distinguished.

Another point concerns the appropriate methods of providing forecasts to participatory allocation committees which include representatives of these marginalized groups. Those actors, likely to be illiterate, only with great difficulty will completely understand the data and concepts presented. Further, this assumes that they will make it to the meetings at all, despite the inexistence of state funds to subsidize the transport of participants from distant areas. Meetings are often dominated by new local elites linked to agribusiness and by local technocrats (Taddei, 2005).

Finally, along these lines it is also worth considering how these groups would fare given the water trading discussed concerning allocation of water to high-valued uses and forecast value. As they lack capital, they are unlikely to compete effectively for water under scarcity conditions (though it is a possibility that they could continue to simply take water, given poor monitoring). Yet, as in the Aguas do Vale experiment and the Jaguaribe-Apodi example cited above, they may be assigned at least implicit water rights per currently ‘typical’ allocation practice. It may be the case that these rights are bundled with land ownership or rental. This could lead them to sell their water (or rent land bundled with water) rights for the year and pursue less vulnerable options. As noted above, such rights could help to reduce the vulnerability of the poor. Whether vulnerability reduction is in fact what would result from such policies needs to be better understood.

5. Conclusion

We hope to have made clear the value of considering the societal setting in detail in order to assess the potential value of hydroclimatologic innovations. Various sociopolitical changes may be necessary for innovations to be effective and to create equitably distributed benefits. Some are ‘top-down’, others are ‘bottom-up’, and a number may be occurring now in Ceará.

A better understanding of the conditions under which forecasts are useful, and for whom, should facilitate public and private assessments of their generation and dissemination. Such assessments should include the comparison of forecasts’ net benefits with the potential gains from other policy measures such as providing incentives for relocation and retraining to enter other sectors, or subsidizing specific non-water intensive activities, and/or the promotion of conservation-focused technologies.

17

Forecast suppliers can also better consider re-targeting their efforts based on such research concerning societal trends. We focused here on the use of streamflow forecasts on a seasonal-to-interannual timescale, following a specific existing hydroclimatological innovation. However, in studying water demands and policies we observed also that medium to long-term development and infrastructure initiatives, already being implemented, could be aided by improved forecasts of multi-year droughts. This would be a distinct climate-forecast output not currently developed.

Finally, our observations suggest that the current participatory ethos could in principle gain from interaction with social science research or at least related techniques. For instance, specific group exercises bringing together actual and potential WAC members with realistic scenarios of the use of stream flow forecasts in Ceará could help relevant agencies to assess for themselves whether this forecast innovation offers society an opportunity that should in fact be seized.

Acknowledgements and Disclaimer

This study is based on research conducted from 2001-2004 as part of an ongoing collaborative project between the International Research Institute for Climate Prediction and the government of Ceará. Additional funding has been provided by the National Oceanic and Atmospheric Agency Office of Global Programs, National Science Foundation Center for Decision Making Under Uncertainty Program grant #NSF SES-0345840, University of Miami’s Center for Ecosystem Science and Policy and the Research Institute for the Study of Man. We are grateful for the help of many at the agencies COGERH, FUNCEME, DNOCS, and SRH in Ceará. We take sole responsibility for the views expressed and any errors of fact. For helpful comments we thank Amy Clement, Marissa Steketee and Kimberly Talikoff.

Notes 1 In 1999, for instance, the average monthly income per person within the rural population of the state was R$ 75.40 (around US$ 35). This is less than half of the national minimum wage (IPLANCE, 2002b p. 136). 2 See for instance the World Bank’s (1996) Participation Sourcebook. Also, from a purely theoretical perspective, Tiebout (1956) shows that local provision can be more efficient than central whether the local provider is public or private. Oates (1972) and Klibanoff and Morduch (1995) suggest that a central agency can take into account spillovers across localities while local providers may have better information about local needs and preferences. Yet a local entity may not have the best technical information and know-how (Bird, 1995). From another point of view, some suggest that whether decentralization leads to better outcomes depends upon who has control over local and central decision making (see, e.g., Bardhan and Mookherjee, 2000a, b; Besley and Coate, 2000; Seabright, 1996). 3 Thus it is not surprising that, at the time of writing, there exists a highly contested plan for diversion of water to Ceará of out-of-state waters from the São Francisco River to supply expected demand over the medium term. 4 Ceará’s population density is about 50 per square kilometer (IPLANCE, 2002b), high for semi-area areas globally. 5 In 1970, the population of the state was 4.35 million, and in 1980 it was 5.3 million (Carvalho, 1988). 6 Carvalho (1988) proposes 150,000 as more realistic. There is evidence that in the height of the crisis, Fortaleza was the site of one thousand deaths per day due to starvation and illnesses (Neves, 2000; Villa, 2000). 7 Brazil became independent from Portugal on September 7, 1822, but remained a monarchy until November 15, 1889, when a military coup initiated the country’s republican era. 8 Later, cloud seeding was also tried (see Finan (1998) for a review of the uses of science to “combat” drought). 9 Brazil’s first reservoir (Cedro, in Quixadá, Ceará) was started in 1886 and its construction took twenty years.

18

10 Companhia de Gestão dos Recursos Hídricos do Ceará, created by state law 12.217/1993. 11 According to Brazilian law (Decreto 24643/1934), all rivers that are located entirely inside the borders of one state are under the jurisdiction of that state, while those rivers that cross state frontiers are under federal jurisdiction. 12 Brazil’s President, Mr. Luis Inácio Lula da Silva, initiated a restructuring of DNOCS by appointing a new head and new directors. In Ceará, FUNCEME, the meteorological agency has been shifted across secretariats. The heads of SRH and COGERH were replaced even though the party of the previous governor remained in power. 13 Average population is about 24 thousand inhabitants per municipality in the hinterland. Average urbanization rate is 53% (IPLANCE, 2002a; without Salgado sub-basin, due to its relative independence from the rest of the valley). 14 Evidence indicates that licenses for shrimp farming are being given without proper requirement of treatment of the water returning to the river from the tanks. This pollution, associated with the pollution generated by shrimp farms along the state’s shoreline, is blamed for devastating the coastal crab fishery. See for instance O Povo, April 4, 2003. 15 The Salgado sub-basin, though part of the Jaguaribe Valley, is not directly connected to the network of reservoirs that “perennialized” the river, and therefore it was left out of the commission that allocates the reservoir waters. 16 In this article we do not address local climate prediction efforts that are dominated by individual ‘profetas’ – rain prophets. Widespread attention is paid to these forecasts that are widely disseminated via the media (for details see O Povo, February 1, 2003, articles “Homens do Tempo”, “Doutor da Natureza”, “Ventos que Falam”, “Mestre em Contemplação”; Diário do Nordeste, January 7, 2003, “Experiência Popular Aponta Bom Inverno”; January 13, 2003, “Polêmica: Profetas Populares x cientistas”; Finan, 1998). 17 The alternative, a dynamical model, typically sequentially couples a hydrological model to a Regional Climate Model (RCM) simulation preserving boundary conditions specified by the General Circulation Model (GCM) (Leung et al., 1999; Nijssen et al., 2001; Yu et al., 2003). Biases in forecast rainfall fields are typically substantial, and require statistical correction. Further, only a smaller number of ensemble members are typically simulated and hence it is difficult to accurately assess the probability distribution of the forecast rainfall from a GCM and consequently the seasonal streamflow. Uncertainty propagation from the sequential coupling of GCM-RCM-hydrologic model (Kyriakidis et al., 2001) may further call into question the adoption of the forecasts for short-term water management. Consequently, statistical models using either ocean observations, other proxies, or GCM forecast fields (Hamlet and Lettenmaier, 1999; Piechota et al., 2001; De Souza and Lall, 2003) may be more effective at this point. 18 The existing committee water allocation process considers 3 to 5 scenarios presented by COGERH based upon current reservoir storage and such a zero-inflow forecast. These scenarios differ in the proposed sectoral allocation of water and illustrate the state of reservoir storage over the coming year given such releases. Providing for human consumption over a two-year cycle is emphasized as a primary concern, followed by meeting high-value needs 19 Local commentary about a potential water market in Ceará includes strong negative reactions by left-wing groups in Ceará (anti-market and anti-water-commodification have recently become a dominant discourses among parties such as the Partido Socialista Brasileiro [(PSB) or Brazilian Socialist Party)] or Partido Comunista do Brazil (Communist Party). For example, in April 2003, the PSB had a TV commercial alerting the population against neo-liberal pro-water-market efforts in the country, without explicitly mentioning any group. Many university students and professors [e.g., Diário do Nordeste, March 30, 2003], and government employees (including at COGERH) oppose a market approach to allocation. Ideological aversion to the commodification of water is exemplified in campaign slogans and graffiti around the state against water-market concepts. 20 COGERH data indicates that rice production consumes about half the water used in the Jaguaribe Valley. 21 This shift did not occur as hoped for, despite some efforts to train farmers and provide credit (though the later suffered from the lack of collection of funds from those receiving water). Low capacity to invest and low literacy are also significant constraints. 22 Another institutional issue for achieving efficient outcomes through trading, distinct from charging and collection of fees, is the length of contracts. The simplest trading involves transfers of water within any given growing seasons. However, the set of decisions that should be taken efficiently includes investing for a longer term. For instance, planting fruit crops creates a demand for stable water supply over years. Also, investing in significant port infrastructure creates a demand for drinking and industrial water supply for decades. Thus, the contractual demands for all decisions to be efficient will not always be easily met. 23 In 2001, at the national level, a resolution of the National Council for Water Resources (resolution 16/2001) created legal space for the transfer of water licenses between users. In 2004 the Ceará Legislative Assembly (Assembléia Legislativa) passed a new water law that allows transfers of licenses (outorgas) between users.

19

24 It is noteworthy that the outorga and tariff system became law in 1992 and is still not fully implemented. It is not unusual in Brazil that laws are made but then not put into practice. See Taddei (2004) for discussion of this point. 25 The state would have to choose between permanent and temporary licenses as well as group versus individual. 26 COGERH presents participants in the allocation meetings with on average six alternatives for water release rates from the major reservoirs, along with a simulation of the decrease in total volume for the upcoming six months for each rate. Estimated consumption of urban demand, including for Fortaleza, and hinterland human and animal consumption, are removed prior to the generation of scenarios for the WACs. Water allocation in the participatory commission is limited to local decisions primarily about agricultural use, that represents a very small fraction of the state’s economy; decisions regarding industrial use, as well as future development of the Ceará’s hydrosystem, and its relation to the projected economic development of the state, is heavily centralized in the government bureaucracy. 27 They visit municipal governments, associations and local organizations around the reservoirs and invite them to choose a representative for participation in water commissions. Generally, it should be said that this unprecedented activity has advanced rapidly in a short period. 28 In this context, it may actually be important to embed analysis of forecast use within COGERH, particularly if COGERH has the interest or capacity to manage sectoral/user group annual reliable contracts, associated transactions and potentially secure insurance to backstop their annual reliability guarantees. 29 For example, COGERH technicians working passionately for the WACs are not independent of their bosses. Our fieldwork finds that there have been discussions among some agency heads concerning a more equitable manner to decide on bulk rural-urban water transfers in the future, something perhaps along the lines of a larger-scale WAC. 30 Due to internal disputes in CONERH over this, the legal status of water committees was changed, and they lost most of their autonomy regarding their internal organization, with the passing of Decreto No. 26.462/2001. Severe restrictions on the participation of civil society institutions were created, as well as increased centralization of municipal representation in the hands of the municipal executive, with total exclusion of the municipal legislative (“vereadores”). This suggests a de facto return to a centralized paradigm that law 11996/1992 tried to combat. 31 In the second week of August of 2002, newspapers reported that the largest producer of the valley, the giant fruit multinational Del Monte, signed an agreement of water delivery directly with DNOCS, without participating in discussions of water allocation (see newspaper Diário do Nordeste, August 14th, 2002). That caused friction between the water committees and the administration in the Apodi irrigation area where Del Monte is installed. However, it seems that these cases are becoming increasingly rare. 32 We are currently surveying households concerning their investments and adjustments to water levels. 33 That said, it is possible that these forecasts lower the variance of the information that farmers commonly receive. During strong El Nino years, the media often publicizes many quite general – and at times conflicting - forecasts about the expected rainy season. A stream flow forecast that is more relevant to user activities and takes into account additional variables can send a clearer signal on supply expectations and thus reduce the public confusion, although whether this is the outcome would depend strongly upon public trust in the forecast supplier. 34 Of course the asymmetry will vary by use, e.g. drinking water may be more asymmetric than agricultural production. In the latter, having no rain one year and more the next could be just as good as the average twice. 35 Without going too far into the details here, as the most important hurdles at this point are probably the political and institutional issues, distinguishing water and crop insurance is worthwhile. Conceptually, insurance for crop failure faces ‘moral hazard’ concerns about farmers’ unobserved levels of effort. Given a payout if the crop fails, a farmer may try less to avoid failure than the insurer would prefer. Insurance for delivery of water does not have a role for farmer effort. Payouts for no water are triggered by rain and the release decisions. 36 While insurance may be priced solely from the entire historical record, it may be that insurers more happily enter the market when year-to-year signals of scarcity exist. It is also possible that cutting-edge analysis of climatic regimes, i.e. forecasts of how temporal averages will shift, would assist insurers. 37 It should be noted that even without forecasts, i.e. with the average forecast every year, a market in such places in the water line could coordinate water investments and demand. However, the more information about water supply is available, the more relevant the demand signal. For instance, when water is low, we want the demands for water or greater or lesser certainty to be revealed. Forecasts help this market do that. 38 Here again investments whose benefits occur over several years raise the issue of water in the future.

20

39 Such discussions are often led, in fact, by local academics in light of limited signals by the official agencies about adopting such innovations. Transfer of ideas from the academy to policy can be slow and of course for any given idea does not always occur. Our point is that in this area there have been consistent signals in a specific direction. 40 Field observations indicate that rainfed farmers may sometimes also work as laborers in the fields of irrigated farmers, though analysis of these type of more complex economic linkages are outside the scope of this paper. 41 Decisions on drought relief measures, such as work programs, seed releases, and welfare packages are made at the State level and are not influenced by streamflow forecasts but instead by more general drought forecasts (e.g., by the Climate Outlook Forum, or ENSO forecasts made by various international groups) often communicated to the users in the water system via the WACs (the bulk of the population attains drought information through the media or municipal government). However, some rainfed groups are represented in WAC via union representatives, and could react to the signal of a forecast of reduced precipitation, for instance by starting to agitate (e.g., with the media) in anticipation of the need for relief funds (see Finan, 2001; Kenny, 2002; Neves, 2000, 2002; Villa, 2000; on historical traditions of protest, including ‘food riots’, to elicit government aid; see Lemos (2003) on strategies developed by state technocrats to deal with related movements in the 1990s). 42 See Diário do Nordeste, July 24th, 2004. 43 Our ongoing surveys seek to document how a wide range of producers respond to changes in water availability.

21

References

Albuquerque Júnior, D.M. de.: 2004, ‘Weaving Traditions: The Invention of the Brazilian Northeast’, Latin American Perspectives, 31(135), 42-61.

Adiku, S.G.K. and Stone, R.C.: 1995, ‘Using the Southern Oscillation Index for improving rainfall prediction and agricultural water management in Ghana’, Agricultural Water Management, 29(1), 85-100.

Arumugam, S., Sharma, A. and Lall, U.: 2003, ‘Water Allocation for multiple uses based on probabilistic reservoir inflow forecasts’, in IAHS-IUGG Proceedings, Soppore, Japan.

Bardhan, P.: 2002, ‘Decentralization of Governance and Development’, Journal of Economic Perspectives, 16(4), 185-205.

Bardhan, P. and Mookherjee, D.: 2000a, ‘Corruption and Decentralization of Infrastructure Delivery in Developing Countries’, Working Paper, University of California, Berkeley.

Bardhan, P. and Mookherjee, D.: 2000b, ‘Decentralizing Anti-Poverty Program Delivery in Developing Countries’, Working Paper, University of California, Berkeley.

Besley, T., and Coate, S.: 2000, ‘Centralized Versus Decentralized Provision of Local Public Goods: A Political Economy Analysis’, Journal of Public Economics, 87(12), 2611-2637.

Bird, R.M.: 1995, ‘Decentralizing Infrastructure: For Good or for Ill?’ in Estache, A., (ed.), Decentralizing Infrastructure: Advantages and Limitations, World Bank Discussion Papers, 290, 22 – 51.

Brown, C., and U. Lall: 2005, Water Allocation and Management in a Changing World: Assessing Strategies for Reliable and Equitable Water Use Responsive to Dynamic Risk Assessment, Mimeo, Columbia University, New York.

Callado, A.: 1960, Os Industriais da Seca e os “Galileus” de Pernambuco,. Civilização Brasileira, Rio de Janeiro.

Callahan, B., Miles, E. and Fluharty, D.: 1999, 'Policy implications of climate forecasts for water resources management in the Pacific Northwest', Policy Sciences, 32,269-293.

Campos, N. and Simpson, L.: 2001, ‘Mercado de Água’, in Campos, N. and Studart, T. (eds.), Gestão de Águas: princípios e práticas. ABRH, Porto Alegre.

Carvalho, O. De.: 1988, A Economia Política do Nordeste – Seca, Irrigação e Desenvolvimento, Associação Brasileira de Irrigação e Drenagem, Campus Brasília, Rio de Janeiro.

Cayan, D.R., Redmond, K.T. and Riddle, L.G.: 1999, ‘ENSO and hydrologic extremes in the western United States’ Journal of Climate 12(9): 2881-2893.

Changnon S.A, and Vonnhame, D.R.: 1986, 'Use of climate predictions to decide a water management problem', Water Resources Bulletin, 22,649-652.

22

Chiew, F., McMahon, T.A., Zhou, S-L. and Piechota, T.C.: 2000, ‘Streamflow variability, seasonal forecasting and water resources systems’, in Hammer, G.L., Nicholls, N. and Mitchell, C. (eds), Applications of Seasonal Climate Forecasting in Agricultural and Natural Ecosystems: The Australian Experience, Kluwer Academic Publishers, Dordrecht, pp. 409-428.

Coelho, J.: 1985, As Secas do Nordeste e a Indústria das Secas, Vozes, Petrópolis.

Comitas, L.: 1973, ‘Occupational Multiplicity in Rural Jamaica’ in Comitas, L. and Lowenthal, D. (eds.), Work and Family Life – West Indian Perspectives, Anchor Books, New York.

Costa, A.C.G., Kottak, C.P. and Prado, R.M.: 1997, ‘The Sociopolitical Context of Participatory Development in Northeastern Brazil’, Human Organization, 56 (2), 138-146.

Couper-Johnston, R.: 2001, El Niño - The Weather Phenomenon That Changed the World, Hodder & Stoughton (Coronet), London, p. 354.

Cunniff, R.: 1975, ‘The Birth of the Drought Industry: Imperial and Provincial Responses to the Great Drought in Northeast Brazil (1877-1880)’, Revista de Ciências Humanas, VI.

Davis, M.: 2001, Late Victorian Holocausts: El Niño Famines and the Making of the Third World, Verso, New York.

De Souza, F. and Lall, U.: 2003, ‘Seasonal to Interannual Ensemble Streamflow Forecasts for Ceara, Brazil: Applications of a Multivariate, Semi-Parametric Algorithm’, Water Resources Research, 39, 1307-1325.

Della Cava, R.: 1970, Miracle at Joaseiro, Columbia University Press, New York.

Diário do Nordeste (2002). August 14th, 2002. ‘Del Monte investe US$ 8 milhões em fruticultura’.

--------------- (2003). January 7, 2003, ‘Experiencia Popular Aponta Bom Inverno’.

--------------- (2003). January 13, 2003, ‘Polêmica: Profetas Populares x cientistas’.

--------------- (2003). March 30, 2003, ‘Entrevista, José Carlos de Araújo, professor e pesquisador - Corrente neoliberal quer a mercantilização da água’.

--------------- (2004). July 24th, 2004, ‘Luta pela água compromete abastecimento’.

Dracup, J. A. and Kahya, E.: 1994, ‘The Relationships between United-States Streamflow and La-Nina Events’, Water Resources Research, 30(7), 2133-2141.

Faber, B.A. and Stedinger, J.R.: 2001, ‘Reservoir Optimization Using Sampling SDP with Ensemble Streamflow Prediction (ESP) Forecasts’, Journal of Hydrology, 249, 113–133.

Faoro, R.: 1984, Os donos do poder: formação do patronato politico brasileiro, Globo, Rio de Janeiro.

23

Finan, T.J.: 1998, ‘Birds’ Nests, Donkey Balls, and El Niño: The Psychology of Drought in Ceará, Northeast Brazil’, paper presented at the Annual Meeting of the American Anthropological Association, Philadelphia.

Finan, T.J.: 2001, ‘Drought and Demagoguery: A Political Ecology of Climate Variability in Northeast Brazil’, paper prepared for the workshop Public Philosophy, Environment, and Social Justice, Carnegie Council on Ethics and International Affairs, New York, NY.

Finan, T.J. and Nelson, D.R.: 2001, ‘Making rain, making roads, making do: public and private adaptations to drought in Ceará, Northeast Brazil’, Climate Research, 19, 97-108.

Gaiser, T., Krol, M.S., Frischkorn, H., and de Araújo, J. (eds.).: 2003, Global Change and Regional Impacts: Water Availability and Vulnerability of Ecosystems and Society in the Semi-Arid Northeast of Brazil, Springer, Berlin.

Garjulli, R.: 2001a, ‘Experiência de Gestão Participativa de Recursos Hídricos: a Caso do Ceará’, in Alves, R.F.F. and de Carvalho, G.B.B. (eds.), Experiências de Gestão dos Recursos Hídricos, MMA/ANA, Brasília.

Garjulli, R.: 2001b, Oficina Temática: Gestão participative dos recursos hídricos – Relatório Final, PROÁGUA/ANA, Aracajú.

Garjulli, R., de Oliveira, J.L.F., da Cunha, M.A.L., de Souza, E.R., and Folhes, M.T.: 2002, Projeto Marca D’Água, Relatórios Preliminares. A Bacia do Rio Jaguaribe, Ceará – 2001, Núcleo de Pesquisa em Políticas Públicas/Projeto Marca D’Água, Brasília.

Girão, R.: 1986, A Marcha do Povoamento do Vale do Jaguaribe (1600-1700), SUDENE/GEVJ, Fortaleza.

Glantz, MH.: 1982, 'Consequences and responsibilities in drought forecasting: The case of Yakima, 1977', Water Resources Research, 18, 3-13.

Goddard, L., Mason, S.J., Zebiak, S.E., Ropelewski, C.F., Basher, R., and Cane, M.A.: 2001, ‘Current approaches to seasonal-to-interannual climate predictions’, International Journal of Climatology, 21(9), 1111-1152.

Governo do Estado do Ceará.: 1992, Lei n° 11.996, de 24 de julho de 1992.

Governo do Estado do Ceará.: 1993, Lei n° 12.217, de 18 de novembro de 1993.

Governo do Estado do Ceará.: 2000 a, Ceará – O Semi-Áridos Cearense e os seus desafios, Secretaria de Planejamento e Coordenação, Fortaleza.

Governo do Estado do Ceará.: 2000b, Ceará – Indicadores Sociais e Demográficos – PNAD 1999, Secretaria de Planejamento e Coordenação / Fundação IPLANCE, Fortaleza.

Governo Federal da República Federativa do Brasil.: 1934, Decreto no 24.643, de 10 de julho de 1934 – ‘Código de Águas’.

24

Governo Federal da República Federativa do Brasil.: 1997, Lei nº 9.433, de 8 de janeiro de 1997 – ‘Lei das Águas’.

Governo Federal da República Federativa do Brasil.: 2000, Lei nº 9.984, de 17 de julho de 2000.

Greenfield, G.M.: 1986, ‘Migrant behavior and elite attitudes: Brazil’s Great Drought, 1877-1879’, The Americas, 43(1), 69-85.

Greenfield, G.M.: 1992, ‘The Great Drought and Elite Discourse in Brazil’, Hispanic American Historical Review, 72(3), 375-400.

Greenfield, G.M.: 2001, The realities of images: imperial Brazil and the Great Drought, American Philosophical Society, Philadelphia.

Guidotti, G.: 2003, Utility of Climate Information Based Reservoir Inflow forecasts for Bulk Sector water allocation, Master Thesis, Earth and Environmental Engineering Department, Columbia University, New York, 66 pp.

Hamlet, A.F. and Lettenmaier, D.P.: 1999, ‘Columbia River streamflow forecasting based on ENSO and PDO climate signals’,. J. Water Res. Plan. Man-Asce, 125(6), 333-341.

Hamlet, A. F., Huppert, D. and Lettenmaier, D.P.: 2002, ‘Economic value of long-lead streamflow forecasts for Columbia River hydropower’, J. Water Res. Plan. Man-Asce, 128(2), 91-101.

Hashimoto, T., Stedinger, J.R. and Loucks, D.P.: 1982, ‘Reliability, Resiliency, and Vulnerability Criteria For Water-Resource System Performance Evaluation’, Water Resources Research, 18(1), 14-20.

Hilton, R.W.: 1981, ‘The Determinants of Information Value: Synthesizing some General Results’, Management Science, 27(1), 57-64.

Höynck, S.: 2003, ‘Structure and Possible Development of the Agricultural Sector of Ceará and Piauí’, in Gaiser, T., Krol, M., Frischkorn, H. and de Araújo, J.C. (eds.), Global Change and Regional Impacts – Water Availability and Vulnerability of Ecosystems and Society in the Semiarid Northeast of Brazil, Springer, Berlin.

IPLANCE – Fundação Instituto de Pesquisa e Informação do Ceará.: 2002a, Índice de Desenvolvimento Municipal 2000, Edições Iplance, Fortaleza.

IPLANCE – Fundação Instituto de Pesquisa e Informação do Ceará.: 2002b, Diagnóstico Social do Ceará, Edições Iplance, Fortaleza.

Katz, R.W. and Murphy, A.H. (eds.), 1997, Economic Value of Weather and Climate Forecasts, Cambridge University Press, New York.

Kemper, K., Gonçalves, J.Y.B., and Bezerra, F.W.B.: 1999, Water allocation and trading in the Cariri region – Ceará, Brazil, World Bank Technical Paper No. 427, Washington, D.C.

25

Kenny, M.L.: 2002, ‘Drought, Clientalism, Fatalism and Fear in Northeast Brazil’, Ethics, Place and Environment, 5(2), 123-134.

Klibanoff, P. and Morduch, J: 1995, ‘Decentralization, Externalities and Efficiency’, Review of Economic Studies, 62(2), 223-247.

Kyriakidis, P.C., Miller, N.L., and Kiml, J.: 2001, ‘Uncertainty propagation of regional climate model precipitation forecasts to ydrologic impact assessment’, Journal of Hydrometeorology 2(2): 140-160.

Lall, U. and Miller, C.W.: 1988, ‘An Optimization Model for Screening Multipurpose Reservoir Systems’, Water Resources Research, 24, 953-968.

Lall, U., Sangoyomi, T., and Barbanel, H.D.I.: 1996, ‘Nonlinear dynamics of the Great Salt Lake: Nonparametric short- term forecasting’, Water Resources Research, 32(4), 975-985.

Lemos, M.C., Nelsom, D., Finan, T., Fox, R., Mayorga, D. and Mayorga, I.: 1999, The social and policy implications of seasonal forecasting: a case study of Ceará, Northeast Brazil. NOAA Final Report, Award # NA76GPO385.

Lemos, M. C., T. Finan, R. Fox, D. Nelson, and J. Tucker.: 2002, ‘The use of seasonal climate forecasting in policymaking: Lessons from Northeast Brazil’, Climatic Change, 55, 479 507.

Lemos, M.C.: 2003, ‘A Tale of Two Polices: Climate Forecasting and Drought Relief in Ceará, Brazil’, Policy Sciences, 36, 101-123.

Leung, L. R., Hamlet, A.F., Lettenmaier, D.P. and Kumar A.: 1999, ‘Simulations of the ENSO hydroclimate signals in the Pacific Northwest Columbia River basin’, Bulletin of the American Meteorological Society, 80, 2313-2329.

Magalhaes, A.R.: 2002, ‘Climate, Society and Public Policy: How Climate Affects Society, How Society and Government Respond to Climate Impacts’ presented at Climate Affairs in Latin America: Climate Issues and Policy Responses, Institute of Latin American Studies, Columbia University, New York, NY.

Medeiros Filho, J. and de Souza, I.: 1988, A Seca do Nordeste: Um Falso Problema – A política de combate às secas antes e depois da SUDENE, Vozes, Petrópolis.

Montenegro, A.F.: 2001, Interpretação do Ceará, Casa de José de Alencar / Programa Editorial, Fortaleza.

Nelson, D.R. and Finan, T.: 2000, ‘The Emergence of a Climate Anthropology in Northeast Brazil’, Practicing Anthropology, 22, 6-10.

Neves, F. de C.: 1995, ‘Curral dos Bárbaros: os campos de concentração no Ceará (1915 e 1932)’ Revista Brasileira de História, 15, 93-122.

Neves, F. de C.: 1998, ‘Economia Moral Versus Moral Economia (Ou: O que é economicamente correto para os pobres?)’, Projeto História, 16, 39-57.

26

Neves, F. de C.: 2000, A Multidão e a História, Relume Dumará, Rio de Janeiro.

Neves, F. de C.: 2002, ‘A seca na história do Ceará’, in Souza, S. de. (ed.), Uma Nova História do Ceará, Edições Demócrito Rocha, Fortaleza.

Neves, F. de C.: 2003, ‘Seca, Estado e Controle Social: as políticas públicas de combate às secas no Ceará’ in Braga, E.M.F. (ed.), América Latina: transformações econômicas e políticas, Editora UFC, Fortaleza.

Nijssen. B, O'Donnel, G.M., Hamlet, A.F. and Lettenmaier, D.P.: 2001, ‘Hydrologic sensitivity of global rivers to climate change’, Climatic Change, 50, 143-175.

O Povo (2003). February 1st, 2003, “Homens do Tempo”

--------------- (2003). February 1st, 2003, “Doutor da Natureza”

--------------- (2003). February 1st, 2003, “Ventos que Falam”

--------------- (2003). February 1st, 2003, “Mestre em Contemplação”

--------------- (2003). April 4, 2003, “Carcinicultura, Acaraú: pescadores denunciam degradação na zona costeira”

Oates, W.: 1972, Fiscal Federalism, Harcourt Brace Jovanovich, New York.

Osgood, D., Pfaff A. and Small, A.A.: 2005, Allocation of an Uncertain Renewable Natural Resource: missing markets, quantity options and climate-based forecasts, Mimeo, Columbia University, New York.

Pagano, T.C., Hartmann, H.C. and Sorooshian, S.: 2001, 'Using climate forecasts for water management: Arizona and the 1997-1998 El Nino', Journal of the American Water Resources Association, 37, 1139-1153.

Parente, F.J.C.: 2000, A Fé e a Razão na Política. Conservadorismo e Modernidade das Elites Cearenses, Universidade Federal do Ceará, Fortaleza.

Parente, F.J.C.: 2002, ‘O Ceará dos “coronéis” (1945 a 1986)’ in Souza, S. de (ed.), Uma nova história do Ceará, Edições Demócrito Rocha, Fortaleza.

Piechota, T.C., Chiew, F.H.S., Dracup, J.A. and McMahon, T.A.: 2001, ‘Development of exceedance probability streamflow forecast’ Journal of Hydrologic Engineering, 6, 20-28.

Prado Júnior, C.: 1989, Formação do Brasil Contemporâneo, Brasiliense, São Paulo.

Sankarasubramanian, A., Lall, U., Sharma, A., and Guidotti, G.: 2003, ‘Utility of Climate Information Based Reservoir Inflow Forecasts in Annual Water Allocation – Ceará Case Study’, NOAA workshop on Insights and Tools for Adaptation:Learning from Climate Variability, November 2003, Washington, DC.

27

Sankarasubramanian, A. and Lall, U.: 2004, ‘Dynamic water allocation framework for multiple uses: Utility of climate forecasts towards short-term water management’, presented at Climate Change: A challenge or a threat for water management, Amsterdam.

Seabright, P.: 1996, ‘Accountability and Decentralization in Government: An Incomplete Contracts Model’, European Economic Review, 40, 61-89.

Sharma, A.: 2000a, ‘Seasonal to interannual rainfall probabilistic forecasts for improved water supply management: Part 1 - A strategy for system predictor identification’, Journal of Hydrology, 239, 232-239.

Sharma, A.: 2000b, ‘Seasonal to interannual rainfall probabilistic forecasts for improved water supply management: Part 3 - A nonparametric probabilistic forecast model’, Journal of Hydrology, 239, 249-258.

Taddei, R.: 2004, ‘Os usos da lei e a vida social da legislação hídrica. Notas e reflexões sobre o caso do Ceará’, Teoria e Pesquisa, 45, ppp-ppp.

Taddei, R.: 2005, ‘Of clouds and streams, prophets and profits: the political semiotics of climate and water in the Brazilian Northeast’. Ph.D. Dissertation, Teachers College, Columbia University, New York.

Taddei, R., Broad, K. and Pfaff, A.: 2005, Integrating Climate and Water Management in Ceará: historical background, social and legal structures and implications. In Climate, Water and Society, editors Francisco de Assis de Souza Filho, Upmanu Lall and Renzo Taddei, New York, Fortaleza: FUNCEME/IRI (in press).

Tendler, J.: 2000, ‘The Economic Wars Between the States’, paper presented at the OECD/State Government of Ceará Meeting on Foreign Direct Investment and Regional Development, Fortaleza.

Tiebout, C.M.: 1956, ‘A Pure Theory of Local Expenditures’, Journal of Political Economy, 64, 416-424.

Villa, M.A.: 2000, Vida e Morte no Sertão, Ática / Instituto Teotônio Vilela, São Paulo.

Yao, H. and Georgakakos, A.: 2001, ‘Assessment of Folsom Lake response to historical and potential future climate scenarios; 2. Reservoir management’, Journal of Hydrology, 249, 176-196.

Yu, L., Weller, R.A. and Liu, W.T.: 2003, ‘Case analysis of a role of ENSO in regulating the generation of westerly wind bursts in the Western Equatorial Pacific’, J. Geophys. Res., 108, 10.1029/ 2002JC001498.

World Bank.: 1996, The World Bank Participation Sourcebook. Washington, D.C.

28

Figures

Figure 1. Estimated bulk water consumption per sector 2003. Sources: COGERH and CAGECE.

* Includes the municipal water distribution system and water consumption by the service sector.

** This is a rough estimate of the amount of water that is classified as "lost in transit," i.e. lost to evaporation, percolation, and non-accounted uses along the valley. It does not account for evaporation in reservoirs.

*** Includes small, medium and large scale irrigated agriculture.

Figure 2. Jaguaribe Valley River & Irrigation System.

Figure 3. Jaguaribe Valley and Metropolitan Hydrological System Overview. Numbers below reservoir names reflect their storage capacity. Adapted from Guidotti (2003).

Figure 4. Types of participants in the water allocation meetings (Year of reference: 2003). Source: COGERH.

Figure 5. Seasonal Rainfall and Water System Demands. Water allocation decision timeline

Figure 6. Demography in irrigated agriculture in Jaguaribe Valley – survey sampling. Sources: DNOCS, SEAGRI, municipal governments, rural workers unions and COGERH database.

Figure 7. Public expression of concern about the distribution of water.

Figure 8. Copy of Reservoir Release Scenarios Presented to Water Allocation Committee