Priorities 2007096

Priority Setting for Research on Aquatic Resources

Abstract In contrast to research on farming systems, research on natural resource systems seldom apply rigorous priority setting techniques, mainly due to difficulties associated with estimating research impact ex ante for such systems. This paper presents a replicable approach to priority setting that addresses these difficulties. Assessment is based on multiple criteria, combining economic surplus analysis with subjective scoring, based on information drawn from a cross-country survey of expert opinion. The approach is applied to the WorldFish Center, a global agricultural research organization focusing on living aquatic resources, both farmed and wild. The exercise demonstrates the relevance of conventional evaluation techniques to fisheries research based on a practical application of its impact pathways.

Keywords: research evaluation, multiple criteria analysis, impact assessment, aquaculture, resource management

1

1. Introduction International agricultural research has made tremendous contributions to global food security (Evenson and Golllin, 2003). Nevertheless, there is a growing emphasis on a more effective and efficient allocation of research resources. This entails prioritizing research activities that are expected to have high impact. While all planning implies some set of priorities, often informal and implicit, here “priority setting” denotes an application of explicit, systematic procedures. Likewise, “impact assessment” denotes a specialized evaluation process, which may feed directly into priority setting. Agricultural systems can be divided broadly into farming systems and natural resource systems; for the latter, human exploitation largely involves harvesting wild stocks. Living aquatic resources are a major example: despite the rapid growth of fish farming in the last few decades, capture fishery still accounted for 62% of global fish production in 2004 (FAO, 2005). Precedents abound for research prioritization covering farmed commodities. However, for research in natural resource systems, few studies attempt to assess impact (Pingali, 2001), let alone undertake comprehensive priority setting. Research on farming systems is well suited to conventional techniques of priority setting. Among these, the favored approach is economic surplus analysis (Alston, Norton, and Pardey, 1995). However, for research on natural resource systems, application of economic surplus analysis faces two problems. The first is valuing the total improvement in productivity of such systems due to research. Contrast this with farming systems, where experiments and field trials, combined with adoption studies, can be used to estimate productivity impacts of new technologies. The second is the integration of non-market criteria in decision-making. These include the impact of

2 research on the poor and the environment. The case of aquatic resources highlights these issues, as fisheries (covering aquaculture and capture systems) are a major source of food and livelihood for the poor in developing countries, but face serious environmental threats (FAO, 2004). This study describes an approach for priority setting on aquatic resources research which addresses these problems. To incorporate productivity increases, we extend the economic surplus approach of Briones et al. (2005) to cover both natural and farmed systems across all developing regions. To incorporate a broader set of benefits and costs, we adjust results from economic surplus analysis, using modifiers for equity, environmental sustainability, and other criteria. These modifiers are obtained also by conventional techniques, such as expert judgment and congruence. The resulting modified economic surplus method represents a multiple criteria assessment of research priorities. The method represents a practical and replicable technique towards setting priorities for research on both farmed and natural resource systems The client organization for this exercise is the WorldFish Center, a member of the Consultative Group on International Agricultural Research (CGIAR). The rest of the paper is organized as follows: Section 2 provides a background and reviews related work on priority setting. Section 3 describes the framework for the current priority setting exercise. Section 4 presents the results. Section 5 concludes. 2. Priority setting and research on aquatic resources 2.1. Elements of priority setting Elements of priority setting that need to be defined from the outset are: units, objective, criteria, indicators, and method. “Units” of research are the set of alternatives or options over which priorities are defined. Categories can be based on type of research problem, system under study, geographic region, or some combination of these.

3 “Objective” is the intended form of the statement of priorities. Some exercises seek to assign percentage fund allocations to research units; others have a more modest aim of simply assigning an ordinal ranking to the research units. “Criteria” refer to the standards of assessment. The three commonly applied criteria are economic efficiency, equity, and environmental sustainability. The efficiency criterion uses economic valuation of costs and benefits of research investment. However, this criterion is indifferent to the distribution of the benefits and costs. The equity criterion corrects this by making net benefits to lower income groups more important. Environmental sustainability incorporates a concern for wider impact on ecosystems, as well as the well being of future generations. The criteria would guide the choice of “indicators” to be used in the assessment. Finally, “method” refers to the technique by which the indicators are evaluated to achieve the objective of the exercise. Conventional methods are described in Alston, Norton, and Pardey (1995). Among the simplest of the available methods is congruence, which treats a measure of size or importance of the research units as an indicator for setting priorities. For example, assessment of crop priorities may be determined by quantity or value of crop harvest. Congruence is a popular method owing to its simplicity and the ease of obtaining data. Used as a gauge of economic efficiency, congruence implicitly treats research impact and cost as approximately uniform across research units. On the other hand, benefit-cost analysis allows research units to exhibit different supply impacts and research costs. Streams of future benefits and costs are converted to their current values through discounting. The method yields familiar measures of project worth, such net

4 present value (NPV) or the benefit-cost ratio (BCR). This method however requires more information about ex ante research impacts and investment. Benefit-cost analysis still assumes fixed prices. Significant productivity improvements would however lead to market-level adjustments, which may either reinforce or offset the initial welfare gains from research. Market adjustment can be addressed by economic surplus analysis, which can also serve as an input to an extended benefit-cost comparison. Here “economic surplus” refers to net benefits from the producer’s side and consumer’s side. Computation of changes in economic surplus requires a baseline supply-demand model for simulating price and quantity adjustments, making it an analytically demanding method. The scoring method meanwhile refers either to a method of multiple criteria analysis, or to a method of assigning ratings based on expert judgment. To distinguish between the two, the former is called “aggregated scoring,” the latter “subjective scoring.” For aggregated scoring, each assessment criterion is assigned an indicator; weights for each criterion are defined, and the weighted average of the indicators (suitably normalized) becomes the aggregate score of a research unit. These aggregated scores serve to rank the units. The weights should ideally reflect the value judgments of the stakeholders of the research organization. Meanwhile, subjective scoring is a method of assigning a numerical rating on an arbitrary scale to a research unit based on a particular criterion using expert judgment.

2.2. Examples International research covering a variety of farming systems has been subject to formal priority setting exercises. Examples are documented for the International Rice Research Institute (IRRI), International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), International Livestock Research Institute (ILRI), International

5 Potato Center (CIP), and the International Maize and Wheat Improvement Center (CIMMYT). Table 1 summarizes their respective exercises, according to the elements listed above. Details are found in Evenson, Herdt, and Hossain (1996); IRRI (2002); Kelley, Ryan, and Patel (1995); Thornton et al. (2000); Randolph et al. (2001); Walker and Collion (1997); Pingali and Pandey (2000). For research on fisheries, which encompass both farming and capture systems, the most recent priority setting exercise was conducted by The WorldFish Center to frame its Strategic Plan (ICLARM, 1999a). Research units are defined by region and aquatic resource systems, the latter being defined as a “zone of convergence of the resources, their aquatic environment, and human users” (ICLARM, 1999a; p. 2). Based on Table 2, resource systems are labeled here as follows: Ponds, Lakes, Rivers, Coasts, and Coral reefs; rounding up the list are soft-bottom shelves and open oceans. Culture systems cover Ponds, small parts of Lakes and Rivers, as well as estuaries and lagoons under Coastal waters. Capture systems cover Lakes, Rivers, Coasts, and all of Coral reefs. Units are also defined by region of the developing world, divided as follows: East Asia (EA), Southeast Asia (SEA), South Asia (SA), SubSaharan Africa (SSA), Latin America, (LA), West Asia and North Africa (WANA), and Small Island Developing States (SIDS). The composition of the regions is shown in Table 3 (which incorporates a distinction between mainland and island components of SEA). Owing to environmental concerns and the strong presence of the private sector, WorldFish does not conduct research to raise productivity in marine and brackishwater aquaculture or capture fishery. Output varies widely across production systems and regions (Table 4). The biggest share of production comes from marine capture, while the smallest is from inland capture. Across the regions, the highest total output by far is from East Asia. This

6 is nearly all from China, the global behemoth in fish consumption and aquaculture production. Far behind is the second largest producer, which is SEA, followed closely by Latin America. Inland capture meanwhile plays a big role in Sub-Saharan Africa and South Asia. Environmental sustainability is a major issue in capture fishery. Production trends show a steady growth in capture output, until a leveling off from the mid-1980s . Given a likely overestimate of catch from China, global catches may have even fallen in the 1990s (Watson and Pauly, 2001). As mentioned earlier, much of the world’s stocks are overfished, with the abundance of many large marine predators having fallen by 90% worldwide, primarily due to fishing pressure (Myers and Worm, 2003). Poverty also varies greatly across the regions. Based on 2000 data, poverty incidence in Sub-Saharan Africa is highest at 53%, followed by South Asia at 42%. Poverty is lowest in East Asia (11%). These overall figures are consistent with FAO (2004) estimates for fishery-dependent communities, namely poverty incidence of 26% for Asia, 16% for Latin America, and as much as 46% for Africa. The objective of the first WorldFish priority setting exercise was to sort resource systems and regions (treated separately) into four levels of priority, i.e. Very High, High, Medium, and Low. These levels are associated with ranges of budget allocation: Very High receives 15-30% of WorldFish Center resources; High priority research absorbs up to 15%; Medium priority research receives only 7.5 to 15%. The criteria for ranking resource systems are: potential benefits of research, ability to utilize outcomes of research, existing or anticipated science potential to answer key problems, and research and adoption capacity of intended recipients. These criteria are subjectively scored by experts, who in turn are guided by information sheets containing a profile of the resource system.

7 Meanwhile to rank regions, subjective scoring of each region was undertaken based on poverty, nutritional need, environmental need, nutritional and cultural importance of aquatic resources and their products, and resource availability. Finally, the summary scores were subjected to group discussion, in which WorldFish scientists finalized the ranking of priority resource systems and regions. The results are: •

Very High priority: Ponds in Asia and Sub-Saharan Africa; Coral reefs in SIDS, SEA, and East Africa.

•

High priority: Rivers in SEA – Mekong, and South Asia; Coasts in Southeast Asia, Sub-Saharan Africa, and SIDS.

•

Medium priority: Lakes, soft-bottom shelves

•

Low priority: open oceans The Plan Annex (ICLARM 1999b) notes the following limitations to the priority

setting analysis. First, the assessment did not directly rank region-resource combinations, thus omitting region-specific risks and constraints for a given resource system. Second, the exercise did not undertake a reliable weighting measure to balance future production opportunities with current production figures. These concerns are addressed in the current exercise, which is outlined in the next Section. 3. Framework and method of the study 3.1. Impact pathway for research on aquatic resources For farming systems (such as aquaculture), the impact pathway of research takes the familiar route of technological change, increased productivity at the farm level, and adoption by farmers. For a natural resource system however the impact pathway of publicly-oriented research is different. The main problem from a social perspective is the common pool property of the resource (Ostrom, Gardner, and Walker, 1994). Harvesting decisions across individuals and over time are interdependent, but resource

8 users tend to ignore this interdependence, resulting in overexploitation. Natural resources management (NRM) is necessary to arrest resource degradation. However NRM is often implemented under considerable ignorance regarding the status and value of resources, as well as the likely impacts of human activity, management actions, and institutional arrangements. Research, both in the natural and social sciences, fills in the knowledge gaps; it therefore traces its impact pathway through NRM. A rough parallel can then be drawn between the two types of research: Farm commodity research generates technologies and recommendations for farmers; farmers adopt recommendations and increase farm productivity. Similarly, NRM research produces recommendations, or an evaluation of management options; this improves the quality of NRM decisions, which leads to an increase in resource productivity (Briones et al., 2005).

3.2. The priority setting framework Based on the foregoing impact pathway, the elements of the framework for priority setting in the WorldFish Center were identified after extensive consultations with its scientists and other stakeholders. The units of assessment are combinations of regions and resource systems. The resource systems are the same as those identified earlier, namely: Ponds, Rivers, Lakes, Coasts, and Coral reefs (except that we omit the lowest priority systems, namely soft-bottom shelves and open oceans.) We also retain the previous regional groupings, except SEA is divided into island and mainland regions, making a total of eight regions (East Asia, Latin America, South Asia, SEA Island, SEA Mainland, Sub-Saharan Africa, SIDS, and WANA). There are forty units in total (= 8 regions x 5 resource systems). To avoid confusion, the term “unit” refers to one of these forty region-resource combinations, while “research unit” refers to the set of research activities that impact on that unit. As a simplification, aquaculture

9 corresponds only to Ponds, and capture to all other systems. This follows the WorldFish practice of excluding technological innovation in brackishwater and marine aquaculture, owing to environmental concerns and the strong presence of private research and extension providers. The objective of the exercise is to rank the forty research units in order of priority. The rankings are grouped by priority level for ease of interpretation, namely: Very High, High, Medium, and Low (respectively ranks 1-10, 11-20, 21-30, and 31-40). The criteria of assessment are economic efficiency, equity, and environmental sustainability. The method involves a modified economic surplus analysis represented in Figure 1. The indicator of the efficiency criterion is the BCR, which is computed using economic surplus analysis. Shifters for natural resource systems and for farming systems are separately estimated; the former through expert judgment, the latter through secondary information from related studies. Shifters are incorporated as shocks to a supply-demand model to compute changes in economic surplus. Expert judgment is also applied for obtaining subjective scores for the modifiers, corresponding to the remaining criteria. The equity indicator is an average of the Human Development Index of the UNDP (2003) and a subjective score. The indicators for environmental sustainability is also a subjective score. These indicators are combined with economic surplus using aggregated scoring, with the following weights: 50% for efficiency, 25% for equity, and 25 % for environmental sustainability. The resulting modified economic surplus indicator assigns an ordinal ranking to the forty research units. [Figure 1 about here]

10 3.3. Data sources The opinion survey to elicit expert judgment was conducted by personal interview using structured questionnaires. It primarily targets experts from the WorldFish Center, given their familiarity with WorldFish research, though a number of external scientists and fisheries managers were also interviewed. The individual approach contrasts with group scoring, common in priority setting exercises, which may be biased by group composition and dynamics. The inclusion of external experts also offsets the possibility of disinterested scoring on the part of internal experts. The expert survey covered 42 respondents, 29 of whom were WorldFish scientists, and the remainder drawn from various institutions in Asia, Africa, North America, and Europe. Responses in each questionnaire were elicited only for those resource systems and regions for which the respondent was prepared to make an educated guess. All the questionnaires were prepared following pre-testing, multidisciplinary consultation, and focus group discussions within WorldFish. The baseline supply-demand model is constructed from the global databases of the FAO, namely Fishstat (value and volume of production and trade), FAOStat – Fish, both Primary and Processed, for fish utilization, processing, and trade. The base year is 2001. The data set of the model is developed along the lines of Delgado et al. (2003). For aquaculture research, information is obtained from ex post impact studies, such as: Ofori and Prein (1996); Gupta et al., (1998); Horstkotte-Wesseler (1999); Lane et al. (2001); Hean and Cacho (2001); Dey et al. (2000); Dey and Prein (2003); Thompson et al. (2003); Strehlow (2004); and ADB (2005). Assorted secondary information (such as on research investments) is used in the remainder of the analysis.

11 3.4. Economic surplus analysis The baseline model structure and numerical method closely follows Dey, Briones, and Ahmed (2005), as well as Briones et al (2005). The former describes the AsiaFish model, while the latter applies an AsiaFish-based model to developing country regions and natural resource systems (excluding aquaculture). The change in economic surplus is obtained from a comparison between a scenario with research, and a baseline or counterfactual scenario without research. Scenarios cover a 20-year horizon, with future values discounted to their present value (at an assumed discount rate of 5%). A comprehensive analysis of renewable resource systems would model the interdependence between current harvest and future productivity by introducing stock dynamics. The impact of NRM would then be felt through controls on current harvest (or perhaps harvest effort), resulting in changes in future supply. Briones (2006) offers a prototypical multi-market model with stock dynamics, but notes major data gaps that constrain its application to actual markets and resource systems. Rather than force the issue here with more ad hoc assumptions, we resort to exogenous projections of productivity in capture supply. Hence the with-and-without scenarios share identical exogenous variable projections, except for supply shifts, which differ according to impact of research on supply productivity, as estimated by expert opinion. The more comprehensive approach can be introduced in future work once the real-world dynamics of stocks in large-scale ecosystems is better understood. A final challenge to numerical specification of the model is imputing price elasticities of supply and demand, and income elasticity of consumption. Again we draw on the elasticities of the AsiaFish-related studies. We also impose additional restrictions to minimize arbitrary bias: namely, that the elasticities be as uniform as possible across regions and resource systems; that the resulting parameters permit ready convergence

12 towards an equilibrium solution over multiple periods; and that model runs yield positive changes in economic surplus from supply expansion. Research-induced supply shift is computed from the initial potential productivity improvement, which then propagates according to a logistic process. For capture systems, the adoption path is calibrated from expert opinion about ceiling adoption rate, and adoption rate in 10 years. (The initial adoption rate is arbitrarily set at 0.05 %.) While opinion was surveyed by resource system and region, we opt to apply the average estimates to all capture systems to minimize bias. The resulting mean adoption in 10 years is 8.7% and the mean ceiling adoption is 17.3%. Such ranges however appear to be sharply inconsistent with actual adoption trends for aquaculture innovations. Hence for Ponds, adoption is pegged at 20% in 10 years and 50% maximum. These are conservative figures, compared to actual adoption patterns for agricultural innovation in general and fish farming in particular. Expert opinion is elicited regarding productivity shifts from NRM as follows: The respondent is asked to posit two scenarios, namely: the “business-as-usual” scenario, and the “best-practice” scenario. The latter pertains to an ideal management regime where targets for biological reference indicators are attained. Research is regarded as necessary to implement the best-practice scenario. The respondent is asked to estimate production trends relative to benchmark output for each scenario. Estimates may vary by time horizon, that is, short, medium, and long term (respectively, years 1-5, years 6-10, and years 11-20). For coral reefs, estimates pertain to future status of coral reef formations. The proxy indicator of status is the share of reef area by risk classification (Low, Medium, and High), based on Bryant et al (1998). The link to fish supply is made by eliciting estimates of reef-related fish production by risk category and region. The resulting

13 trends are applied to the estimated reef-related marine capture supply, which is obtained by disaggregating marine capture and identifying, in consultation with biologists, the species groups most closely associated with coral reefs. As for the cost of R&D investment, an estimate is made based on some proportion of the value of production for the relevant unit. For Ponds, the proportion used is 1%. For capture systems the cost ratio to production value is conservatively set at 5%. This is furthermore adjusted downward by applying the percentage only to the adoption domain of best-practice management. The basis of these estimates is as follows: in farming systems, the percent of agricultural value added that research and extension funding should target is often stated at 1% as a rule of thumb (McIntire, 1998). For capture systems, cost estimates are sparser. Balmford et al. (2004) estimate that adequate protection of coastal and coral reef systems would entail a cost of 5 to 18% of the world’s annual value of production. Similarly in developed countries, research-management expenditures range from 5 -30% of the value of the catch (Arnason, 2000). With few exceptions, this is consistent with data from OECD countries for 1999 (OECD, 2003). The high end of the percentage agrees with other estimates of the cost of coral reef protection; for example, the estimated cost of protecting a marine sanctuary in the Philippines is US$ 21,000/km2/yr, which is 50% of the value of sustainable fish production (White, Ross, and Flores, 2000). We take the lower end of the estimates to avoid unduly penalizing NRM research.

3.5. Modifiers Modifiers were taken from subjective scoring, obtained through questionnaires on environmental sustainability and equity. All scores are defined on a scale of 1 to 5, with 5 being the most favorable to the priority rank of a research unit. For

14 environmental sustainability, each of the forty research units is rated through five subcriteria as follows (weights given in parenthesis): •

Importance of the unit to global stock or biomass of aquatic organisms (10%)

•

Importance of the unit to global biodiversity (10%)

•

Vulnerability of the unit to human activity (20%)

•

Potential contribution of the research unit to restoring aquatic habitats (30%)

•

Potential contribution of the research unit to promoting sustainable technologies and practices (30%) For equity, ratings are elicited regarding severity of poverty (by unit). Here

“poverty” takes on a multi-dimensional character, incorporating such concerns as access and exclusion, which are difficult to quantify using conventional poverty measures. As mentioned earlier, the equity score incorporates the HDI indicator. A regional HDI is computed as the population-weighted average of the country HDIs. A resource system HDI per region can be computed by apportioning the regional HDI, using the relative percentages of the resource system ratings by region, obtained from the subjective scoring for equity. 3.6. Sensitivity analysis Sensitivity analysis measures the difference between a given ranking with a comparison ranking, where a comparison ranking is obtained by modifying an assumption or set of assumptions used to derive the given ranking. We say that a given ranking is not “robust” to the change in assumptions if there are large differences between a given and comparison ranking. Differences are gauged using several measures. The first set of measures evaluate the two rankings over all the units, and are composed of the following: the correlation coefficient or COR; the mean absolute deviation or MAD (the average discrepancy in ranking); and the root mean square

15 deviation, or RMSD (an average measure where bigger rank discrepancies are weighted more heavily). To define the last two measures, label the units arbitrarily by the by u = 1, 2, …, 40; let G (u ) be a function that assigns the rank of u based on the given

ranking, C (u ) be similar function for the comparison ranking; let abs be the absolute value function . Then: 40

MAD = (1 40 ) ∑ abs[G (u ) − C (u )] u =1

RMSD =

40

(1 40 ) ∑ [G (u ) − C (u )]

2

u =1

Another set of measures makes comparisons only over a subset of the units. Here the subsets are based on the priority levels Very High, High, Medium, and Low, for which we respectively assign a numerical index i = 1, 2, 3, 4. Define an integer function LG (u ) , which assigns to u its priority level i, based on the given ranking; similarly we define a category function LC (u ) for the comparison ranking. Finally, let

U i = {u : LG (u ) = i} , i.e. U i is the set of units in priority level i. The mean absolute level deviation for level i or MLDi , is computed as follows: MLDi =

1 ∑ abs [ LG(u ) − LC (u )] 10 u∈Ui

That is, MLDi measures the degree to which the priority level of a unit based on a given ranking is matched under a comparison ranking; the lower the MLDi , the closer the match.

16

4. Results 4.1. Assessment of economic efficiency Table 5 presents the results from the assessment of economic efficiency. The value of production (obtained from the base data set), which serves as the congruence indicator, corresponds closely with the production quantities in Table 4. Coasts are the largest units overall, namely in Latin America, East Asia, and SEA Mainland. Coasts also tends to be the largest system within each region. The exception is SIDS, where the whole of marine capture is imputed to Coral reefs owing to the relative abundance of this resource, making it the biggest system in that region. The second largest resource system in each region is Ponds, except for Sub-Saharan Africa, where the second biggest is both Lakes and Rivers. Coral reefs is the smallest resource system in every region except SEA Island and SIDS. The annual supply shift attributed to research varies greatly by resource system. The biggest shifts are for Coral reefs; expert opinion rated WANA and South Asia as having the highest potential NRM impact. In the other capture systems the supply shifts are relatively small; productivity improvement is likewise conservative for Ponds (slightly below one percent gain per annum). Starting value and annual shift together explain a great deal of the change in economic surplus (shown as a discounted value). Large changes in surplus are associated with large units and large shifts. In most of the Coral reefs however the large shifts cannot offset the small size of the units, leading to small changes in economic surplus. Note however that large systems are not necessarily those which exhibit a greater change in surplus, as first and second order effects may be greater for a given proportional shift in supply. This appears to hold for Coasts, especially in Latin America. Remarkably, change in economic surplus is highest for Ponds, despite their

17 middle-rank based on congruence. A ranking based on economic surplus (not shown in the Table) would classify East Asia, Latin America, South Asia, and WANA as Very High priority. A further adjustment arises from introducing cost of research investment and dissemination. The BCRs are well within the range encountered in the literature (see e.g. Alston et al, 2000). The importance given to Ponds is enhanced further; only SIDS fails to make it to the top ten ranking units based on BCR. The congruence factor for Coasts is now doubly offset by the high cost and small economic surplus impact; hence Coasts tend to be the lowest ranking units. In between are the Lakes and Rivers; Coral reefs are somewhat mixed. Note that the cost adjustment allows many of the Coral systems to overcome the inherent size disadvantage – an effect most noticeable in South Asia and Sub-Saharan Africa.

4.2. Sensitivity analysis The foregoing discussion has noted the variability of the mean supply shift estimates by region and resource system. For just the capture systems, the coefficient of variation for expert opinion on research shifters (by unit and time horizon) range from 1.3 to 13.7, highlighting the need for sensitivity analysis regarding these shifts. The rankings to be compared are derived from following indicators: congruence (which sets all shifts to zero); economic surplus analysis or ESA (which controls for differences in cost), BCR (which sets the original ranking for this exercise); ESA and BCR with greater supply shifts (respectively, ESA-High and BCR-High); and ESA and BCR using smaller supply shifts (respectively, ESA-Low and BCR-Low). Sensitivity tests are conducted only for the capture supply shifts. The High-shift comparisons use one standard deviation of the individual responses to the expert opinion survey. In the case of Coral reefs, the standard deviation is imputed from the

18 adjustments made for the Coastal shifters of the same region. As the standard deviation typically exceeds the mean; the Low shift comparisons apply a zero productivity change (within the adoption domain of NRM). The given rankings are ESA and BCR, to be compared with the all the alternative rankings. The results for COR, MSD, and RMSD are shown in the Table 6. For the first three columns the given ranking is ESA, and for the next three is BCR. There is a very high degree of correlation between the three ESA-based rankings (above 0.9); correlation with congruence is lower (just below 0.8), but still high. In contrast, the BCR-based rankings show low (to slightly negative) correlation with the ESA-based rankings. The BCR-based rankings are correlated with one another; however, as these are ratios, the strength of correlation is weaker than the cross-correlation across the ESA and congruence rankings. The same patterns are observed in the MAD and RMSD measures. The MLDi measures are shown in Table 7. The given rankings for the top seven rows is ESA, while that of the bottom seven rows is BCR. Patterns observed in the Table closely parallel those observed in Table 6. ESA-based rankings are most closely matched, followed ESA and Congruence. There is a tight fit across similar rankings; for instance, units rated Very High under ESA are rated less than a third of a level lower on average under Congruence. However there is a relatively sharp mismatch between BCR-based rankings and ESA- or Congruence rankings. The BCR-based rankings are most closely matched to each other (but the fit is not as tight as among the ESA-based rankings). Overall the sensitivity analysis fails to detect instability of BCR rankings or priority classifications when estimated research shifters are varied.

19 4.3. Modifiers The environmental scores and ranking shown in Table 8. The scores are reported as normalized figures (relative to the maximum score); the third column displays the actual deviations from the economic efficiency (BCR) ranking (a negative value implies that the unit ranks better under the environmental criterion than under the efficiency criterion). Coasts are generally rated as Very High priority, accounting for six out of the top ten positions. This is understandable given the size of marine fish stocks, although Coasts in Latin America and East Asia fall under lower priority levels. Coral reefs in Southeast Asia Island and in SIDS are also Very High priority. A couple of inland systems (Lakes and Rivers from East Asia), round up the environment top scorers. This pattern is somewhat the reverse of the efficiency ranking of these units, as shown in the negative deviations. Ponds in general are rated Low priority as these are artificial rather than natural systems. Size also matters for the Low priority level, particularly for Coral in East Asia, as well as Rivers in SIDS and WANA. Equity scores (also normalized) and rankings follows in Table 9. Grouping the units by region clearly shows the overall patterns in the scores: Sub-Saharan Africa units are all top-ranked, as are all the South Asia units (except Ponds). The only other top-ranked unit is Rivers in SEA Mainland. At the other extreme, Latin America units are rated Low priority (except for Ponds). Rankings for South Asia and SEA-Mainland units improve under the equity criterion, compared to their ranking under the efficiency criterion.

4.4. Overall rankings Final rankings based on aggregated scores are shown in Table 10. The last three columns report deviation of final rank from the ranking based on a single criterion (i.e. efficiency alone, equity alone, and environmental sustainability alone). Given the high

20 weight for efficiency (50%), most of the earlier classifications of priority level under efficiency are preserved. Ponds occupy six out of the top ten positions; the remainder is made up by Coral reefs. Sub-Saharan Africa and South Asia are prominent in the Very High priority level, despite the low weight (25%) given to equity. For these regions there is considerable overlap between efficiency and equity. Units under Low priority consist of small-sized resource systems, found in SIDS, WANA, SEA Island, and Latin America. However size is not all that matters, as Coasts in East Asia and Latin America fall under Low priority given their low efficiency scores and mediocre scores under the other criteria. The middle levels (High and Medium) consist of a heterogeneous mix of regions and resource systems. The final research priorities are not completely divergent from the existing strategic priorities. For example, there is a similar emphasis on Ponds, in Sub-Saharan Africa and Asia. Also consistent is the secondary importance given to inland waters and coastal systems, as well as to regions such as Latin America and WANA. The high environment scores for Coral reefs in SIDS and SEA Island also keeps them in the highest priority level. There are however areas of contrast: other Ponds (in East Asia and WANA) are given emphasis, as are Coral reefs in South Asia and Sub-Saharan Africa.

5. Conclusion Priority setting is widely practiced in international agricultural research as a tool for informing investment choices. Established methods include economic surplus analysis, congruence, and subjective scoring. Analysis may be based on multiple criteria to take into account impacts on the economy, on the poor, and on the environment. However, unlike for farming systems, research in natural resource systems has seldom been subjected to these methods, as estimation of research impact has been problematic.

21 For such systems, evaluation of priorities is often based on informal judgment and casual analysis. This study conducts a modified economic surplus analysis for research on both aquaculture and capture systems in fisheries, following a well-defined impact pathway. Modifiers corresponding to other assessment criteria adjust the economic surplus indicator. The approach is applied to international research on living aquatic resources conducted by the WorldFish Center. Broad patterns in the research priorities identified here appear to be robust, based on sensitivity analysis. Compared with existing strategic priorities of the Center (derived from subjective judgment), a much more comprehensive assessment is made possible by modified economic surplus. Some of the earlier priorities were maintained, such as an emphasis on poor developing regions in South Asia and Sub-Saharan Africa; this highlights the consistency rather than trade-off between efficiency and other criteria for these regions. On the other hand, some new results emerge, such as a stronger and wider emphasis on freshwater aquaculture and coral reef systems. This study demonstrates the flexibility of conventional techniques in research prioritization over all types of agricultural systems. Such techniques are not intended to replace the role of the decision-maker’s judgment in the practice of research management. It does however provide a more transparent, replicable, and empirical basis for research planning.

Acknowledgements This study was funded by the WorldFish Center. The views expressed in this paper are the authors’ and not of any organization. The authors are grateful to Chen Oai Li for research assistance; to the survey respondents, for sharing their time and

22 expertise; and to the editor and anonymous referees, whose substantive comments have considerably improved the paper. The usual disclaimer applies.

References Alston, J., Chan-Kang, C., Marra, M., Pardey, P., Wyatt, T., 2000. A meta-analysis of rates of return to agricultural R & D: ex pede herculem? Research Report No. 113. International Food Policy Research Institute, Washington, D.C. Alston, J., Norton, G., Pardey, P., 1995. Science under Scarcity: Principles and Practice for Agricultural Research Evaluation and Priority Setting. Cornell University Press, Ithaca. Arnason, R., 2000. Costs of fisheries management: theoretical and practical implications, in: Proceedings of the 2000 Biennial Conference of the International Institute for Fisheries Economics and Trade. Asian Development Bank, 2005. An impact evaluation of the development of Genetically Improved Farmed Tilapia and their dissemination in selected countries. Asian Development Bank, Manila. Balmford, A., Gravestock, P., Hockley, N., McClean, C., Roberts, C., 2004. The worldwide costs of marine protected areas. Proceedings of the National Academy of Sciences 101, 9694-9697. Briones, R. 2006. ”Projecting Future Fish Supplies Using Stock Dynamics and Demand.” Fish and Fisheries 7(4):303-315. Briones, R., Dey, M., Stobutzki, I., Prein, M., 2005. Ex ante impact assessment for research on natural resources management: methods and application to aquatic resource systems. Research Evaluation 14: 217-227. Bryant, D., Burke, L., McManus, J., Spalding, M., Ablan, M., Victor Barber, C., Cabote, C., Cesar, H., Done, T., Gorospe, M., Guzman, H., Hallock, P.,

23 Hawkins, J., Hayman, A., Hodgson, G., Jameson, S., Maragos, J., McAllister, D., Meñez, L., Ming, C-L., Moola, S., Muthiga, N., Reyes, K., Roberts, C., Schueler, F., Uy, I., Vergara, S., White, A., Wilkinson, C., 1998. Reefs at Risk: A Map-based Indicator of Threats to the World’s Coral Reefs. World Resources Institute, ICLARM, World Conservation Monitoring Centre, and United Nations Environment Programme. Delgado, C., Wada, N., Rosegrant, M., Meijer, S., Ahmed, M., 2002 Fish to 2020: Supply and Demand in Changing Global Markets. International Food Policy Research Institute, Washington, and WorldFish Center, Penang. Dey, M., Briones, R., Ahmed, M., 2006. Disaggregated Projections of Fish Supply, Demand, and Trade: Baseline Model and Estimation Strategy. Aquaculture Economics and Management 9, 113-139. Dey, M., Eknath, A., Sifa, L., Hussain, M., Thien, T., Van Hao, N., Aypa, S., Pongthana, N., 2000. Performance and Nature of Genetically Improved Farmed Tilapia: A Bioeconomic Analysis. Aquaculture Economics and Management 4, 83-106. Dey, M., Prein, M., 2003. Participatory research at landscape level: floodprone ecosystems in Bangladesh and Vietnam, in: Pound, S., Snapp, S., McDougall, C., Braun A.(Eds.), Managing Natural Resources for Sustainable Livelihoods: Uniting Science and Participation. Earthscan and IDRC, London, pp. 223-225. Evenson, R., Gollin, D., 2003. Assessing the Impact of the Green Revolution, 1960 to 2000. Science 300:758-762. Evenson, R., Herdt, R., Hossain, M., 1996. Priorities for Rice Research: Introduction, in: Evenson, R., Herdt, R, Hossain, M. (Eds.), Rice Research in Asia: Progress

24 and Priorities. CABI International and International Rice Research Institute: Wallingford, pp. 3-16. Food and Agriculture Organization [FAO], 2004. State of the World Fisheries and Aquaculture. FAO, Rome. FAO, 2005. Fishstat.. (Downloaded 1 June 2005). Gupta, M., Sollows, J., Mazid, M., Rahman, A., Hussain, M., Dey, M., 1998. Integrating Aquaculture with Rice Farming in Bangladesh: Feasibility and Economic Viability, Its Adoption and Impact. Technical Report 55. ICLARM, Manila. Hean, R., and O. Cacho, 2002. Mariculture of giant clams, Tridacna crocea and T. derasa: management for maximum profit by smallholders in Solomon Islands. Aquaculture Economics and Management 6, 373-395. Horstkotte-Wesseler, G., 1999. Socioeconomics of rice-aquaculture and IPM in the Philippines: Synergies, Potential, and Problems. Technical Report 57, ICLARM, Manila. ICLARM, 1999a. ICLARM Strategic Plan 2000-2020. Contribution 1544: ICLARM, Manila. ICLARM, 1999b. Aquatic Resources Research in Developing Countries: Data and Evaluation by Region and Resource System. Supplement to the ICLARM Strategic Plan 2000-2020. Working Document 4, ICLARM, Manila. IRRI, 1999. Sustaining Food Security Beyond the Year 2000: A Global Partnership for Rice Research. Medium-Term Plan 2000-2002. IRRI, Makati, Philippines. Kelly, T., J. Ryan, and B. Patel, 1995. Applied participatory priority setting in international agricultural research: making trade-offs transparent and explicit. Agricultural Systems, 49, 177-216.

25 Lane, I., Oengpepa, C., Bell, J., 2001. Production and grow-out of the black-lip pearl oyster Pinctada margaritifera. Aquaculture Asia 3, 5-7. McIntire, J., 1998. Coping with Fiscal Stress in Developing-Country Agricultural Research, in: S. Tabor, W. Janssen, and H. Bruneau (Eds.), Financing Agricultural Research: A Sourcebook. The Hague: International Service for National Agricultural Research (ISNAR), pp. 81-96. Myers, R., Worm, B., 2003. Rapid worldwide depletion of predatory fish communities. Nature 423:280-283. OECD, 2003. The Costs of Managing Fisheries. Organization for Economic Cooperation and Development. http://www1.oecd.org/publications/ebook/5303011E.PDF (accessed January 2004). Ofori, J. Prein, M., 1996. Rapid Appraisal of Low-Input Aquaculture Systems, in: Prein, M., Ofori, J., Lightfoot, C. (Eds.), Research for the Future Development of Aquaculture in Ghana. ICLARM, Manila. Ostrom, E., Gardner, R., Walker, J., 1994. Rules, Games and Common-pool Resources. Westview Press, Boulder, Colorado. Pingali, P., Pandey, S., 2000. Meeting World Maize Needs: Technological Opportunities and Priorities for the Public Sector, in: World Maize Facts and Trends 1999-2000. Centro Internacional de Mejoramiento de Maiz y Trigo (CIMMYT), Mexico City. Pingali, P., 2001. Milestones in impact assessment research in the CGIAR, 1970-1999. With an annotated bibliography of impact assessment studies conducted in the CGIAR, 1970-1999, prepared by Matthew P. Feldmann. Mexico, D.F.: Standing Panel on Impact Assessment, Technical Advisory Committee, CGIAR.

26 Rajasuriya, A., Zahir, H., Venkataraman, K., Islam, Z., Tamelander, J., 2004. Status of Coral Reefs in South Asia: Bangladesh, Chagos, India, Maldives, Sri Lanka, in: Wilkins, C. (Ed.) Status of Coral Reefs of the World: 2004. Australian Institute of Marine Science, Queensland, pp. 213-234. Randolph, T., Kristjanson, P., Omamo, S., Odero, A, Thornton, P., Reid, R., Robinson, T., Ryan, J., 2001. A framework for priority setting in international livestock research. Research Evaluation 10, 142-160. Strehlow, H., 2004. Economics and management strategies for restocking sandfish in Vietnam. NAGA-The WorldFish Center Quarterly 27, 36-40. Thompson, P., P. Sultana, and A. Khan, 2003. Aquaculture extension impacts in Bangladesh: a case study from Kapasia, Gazipur. Technical Report 63. WorldFish Center, Penang. Thornton, P., T. Randolph, P. Kristjanson, W. Omamo, A. Odero, J. Ryan, 2000. Assessment of Priorities to 2010 for the Poor and the Environment. Impact Assessment Series No. 6, ILRI, Nairobi. UNDP, 2003. Human Development Report 2003: Millennium Development Goals: A compact among nations to end human poverty. UNDP, New York. Walker, T., and M. Collion, 1997. Incorporating poverty in priority setting: the CIP’s 1998-2000 Medium Term Plan. Program Report 1997-1999. CIP, Lima. Watson, R., Pauly, D., 2001. Systematic distortions in world fisheries catch trends. Nature 414, 534-536. White, A., M. Ross, and M. Flores, 2000. Benefits and costs of coral reef and wetland management, Olango Island, Philippines, in Cesar, H. (Ed.), Collected Essays on the Economics of Coral Reefs. CORDIO, Sweden, pp. 215 – 227.

27

List of Figures Figure 1: Schematic diagram of the priority setting process

List of Tables Table 1: Summary of priority setting elements in selected international agricultural research centers Table 2: Resource system definitions from ICLARM’s Strategic Plan Table 3: Country composition of the regions Table 4: Fish production in 2001, by production system and region, in ‘000 mt Table 5: Unit indicators and ranking based on economic efficiency Table 6: Comparisons across priority rankings by economic efficiency indicator Table 7: Comparisons of priority categories by economic efficiency indicator Table 8: Unit indicators and ranking based on environment scores Table 9: Unit indicators and ranking based on equity scores Table 10: Final scores and rankings of units

28

Table 1: Summary of priority setting elements in selected international agricultural research centers Center IRRI

Units Regions Ecosystems

Objective Allocate budget shares by unit

Criteria Supply requirement Equity Sustainability Presence of alternative research supplier

ICRISAT

Themes

Ordinal ranking; Cumulative investment requirement

Efficiency Equity Internationality Sustainability

ILRI

Themes

Ordinal ranking; Cumulative investment requirement

Efficiency Equity Internationality Sustainability Capacity-building

CIP

Projects

Ordinal ranking

CIMMYT

Constraint x region x ecology

Ordinal ranking

Efficiency Poverty Efficiency Poverty Presence of alternative research supplier

Sources: See references cited in the text.

Indicators Projected demand; yield gap Per capita calorie deprivation Rice area under unfavorable environment (%) National agricultural research spending NPV of production increase (with modifiers) Poverty incidence and female illiteracy rate Simpson index Subjective scoring NPV of production increase (with modifiers) Poverty index Simpson index Subjective scoring Subjective scoring

Method Aggregated scoring with equal weights; modifiers include the gender development index and yield gap

NPV of production increase Poverty index Production index Poverty incidence Extent of subsistence farming

Benefit-cost analysis Aggregate scoring Aggregate scoring: 50% efficiency, 30% poverty, 20% alternative supplier

Benefit-cost analysis Aggregate scoring with equal weights

Benefit-cost analysis Aggregate scoring: 30% efficiency, 25% poverty, 20% sustainability, 15% capacity-building, 10% internationality

29

Table 2: Resource system definitions from ICLARM’s Strategic Plan Resource

Description

Ponds

Small freshwater bodies, usually artificial, where aquaculture is possible. Includes flooded fields where aquaculture is integrated with agriculture.

Lakes

Primarily freshwater, includes reservoirs, and small water bodies. Lakes are natural waterbodies. Reservoirs are natural or artificial waterbodies primarily used for irrigation, power generation, and household water supply. Small water bodies have a surface less than 10 km2.

Rivers

Includes floodplains and streams. Streams and rivers are flowing waters, while floodplains are low-lying areas adjacent to watercourses, subject to periodic or near-permanent inundation.

Coasts

Coastal waters include estuaries and lagoons, and critical habitats, such as mangroves. Coastal waters extend up to 10 m in depth, encompassing most fishing grounds of small-scale fishers.

Coral

Coral reefs refer to continental and island shelves in tropical oceans in which reef-building corals are dominant features.

Source: ICLARM (1999a)

30

Table 3: Country composition of the regions Regions East Asia (EA)

Countries China, Mongolia, North Korea

Latin America (LA)

Argentina, Belize, Bolivia, Brazil, Colombia, Chile, Costa Rica, Ecuador, El Salvador, Guatemala, Guyana, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Suriname, Uruguay, Venezuela

South Asia (SA)

Bangladesh, Bhutan, India, Nepal, Pakistan, Sri Lanka

Southeast Asia Island (SEA Island) Southeast Asia Mainland (SEA Mainland) Small Island Developing States (SIDS)

Indonesia, Malaysia, Philippines Cambodia, Laos, Myanmar, Thailand, Vietnam Antigua and Barbuda, Bahamas, Barbados, Cape Verde, Comoros, Cook Islands, Cuba, Dominica, Dominican Republic, Fiji Islands, Grenada, Guadeloupe, Jamaica, Kiribati, Maldives, Martinique, Mauritius, Micronesia, Nauru, New Caledonia, Niue, Northern Marianas Islands, Palau, Papua New Guinea, St. Helena, St. Kitts and Nevis, Sta. Lucia, Samoa, SaoTome and Principe, Seychelles, Solomon Islands, Timor- Leste, Tokelau, Tonga, Trinidad and Tobago, Tuvalu, Vanuatu, Wallis and Futuna Islands

Sub-Saharan Africa (SSA)

Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Congo, Democratic Republic of, Congo, Republic of, Cote d’Ivoire, Djibouti, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mozambique, Namibia, Niger, Nigeria, Rwanda, Reunion, Senegal, Sierra Leone, Somalia, Sudan, Swaziland, Tanzania, Togo, Uganda, Western Sahara, Zambia, Zimbabwe

West Asia and North Africa (WANA)

Afghanistan, Algeria, Bahrain, Egypt, Iran, Iraq, Jordan, Kuwait, Lebanon, Morocco, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia, United Arab Emirates, Yemen, Oman, Turkey, Libyan Arab Republic

31

Table 4: Fish production in 2001, by production system and region, in ‘000 mt Inland capture

East Asia

Marine capture

Freshwater culture

Brackishwater and

TOTAL

marine culture

(by region)

2,293

14,813

16,979

16,641

50,725

515

16,251

8,778

3,130

20,774

South Asia

1,758

3,995

2,773

956

9,481

Southeast Asia

1,449

12,595

2,313

4,730

21,087

19

512

17

33

581

1,719

3,413

51

53

5,236

429

2,885

386

841

4,540

8,181

54,463

23,396

26,385

112,425

Latin America

SIDS Sub-Saharan Africa WANA

TOTAL (by system)

Source: Authors’ calculations based on FAO (2005).

32

Table 5: Unit indicators and ranking based on economic efficiency System Region

Coasts

Coral reefs

Lakes

Ponds

Rivers

EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA

Value of Output ($ ‘000) 18,733 24,876 10,502 12,901 19,226 0 7,702 5,340 19 131 56 2,431 226 2,242 747 405 704 923 1,521 672 1,612 30 1,680 868 13,622 2,419 2,750 2,467 2,060 168 144 1,048 347 455 749 331 794 15 1,680 427

Source: Authors’ calculations

Annual supply shift (%) 0.41 0.44 0.59 0.32 0.42 0.42 1.06 0.83 2.15 2.15 5.69 2.15 2.15 2.15 6.44 2.49 0.81 0.97 0.81 1.13 0.82 0.58 0.94 0.65 0.98 0.98 0.98 0.98 0.98 0.98 0.98 0.98 0.81 0.97 0.81 1.13 0.82 0.58 0.94 0.65

Change in surplus ($‘000) 900 849 1,092 1,133 1,157 0 883 1,718 3 23 207 819 42 379 361 260 177 91 106 36 103 2 79 34 9,185 1,473 3,393 1,462 1,222 164 118 918 86 51 52 18 50 1 79 17

BenefitCost ratio

Rank

0.77 0.15 1.23 0.90 0.63 0.00 1.35 1.55 2.87 0.76 6.30 2.74 1.89 2.97 5.11 2.76 1.11 2.21 1.14 1.15 1.03 1.38 0.85 0.70 4.72 4.38 7.36 4.17 3.12 2.55 4.66 5.29 1.12 2.51 1.14 1.15 1.03 1.25 0.85 0.70

34 39 22 31 38 40 20 18 11 35 2 13 17 10 4 12 28 16 26 24 30 19 32 37 5 7 1 8 9 14 6 3 27 15 25 23 29 21 33 36

Priority Level Low Low Medium Low Low Low High High High Low Very High High High Very High Very High High Medium High Medium Medium Medium High Low Low Very High Very High Very High Very High Very High High Very High Very High Medium High Medium Medium Medium Medium Low Low

33

Table 6: Comparisons across priority rankings by economic efficiency indicator Indicator

Correlation (ESA)

MAD (ESA)

RMSD (ESA)

Correlation (BCR)

MAD (BCR)

RMSD (BCR)

Congruence

0.79

6.2

7.4

-0.06

13.2

16.8

ESA

1.00

0.0

0.0

0.42

9.6

12.5

BCR

0.42

9.6

12.5

1.00

0.0

0.0

ESA – High

0.94

2.7

4.0

0.29

10.8

13.7

BCR – High

0.05

12.6

15.9

0.80

6.1

7.3

ESA – Low

0.93

2.8

4.3

0.31

10.3

13.5

BCR – Low

0.33

10.6

13.4

0.59

7.5

10.4

Source: Authors’ calculations

Table 7: Comparisons of priority categories by economic efficiency indicator Indicator

Very High

High

Medium

Low

Congruence

0.30

1.00

0.60

0.30

ESA

0.00

0.00

0.00

0.00

BCR

0.90

0.80

0.60

0.90

ESA – High

0.40

0.50

0.30

0.20

BCR – High

1.50

1.20

0.50

1.40

ESA – Low

0.10

0.40

0.30

0.00

BCR – Low

0.90

1.10

1.10

0.70

Congruence

1.20

1.20

0.60

1.80

ESA

0.50

0.90

0.60

1.20

BCR

0.00

0.00

0.00

0.00

ESA – High

0.90

0.90

0.70

1.50

BCR – High

0.50

0.90

0.40

0.40

ESA – Low

0.60

0.80

0.70

1.30

BCR – Low

0.60

1.10

0.70

0.40

Relative to ESA

Relative to BCR

Source: Authors’ calculations

34

Table 8: Unit indicators and ranking based on environment scores System

Coasts

Coral reefs

Lakes

Ponds

Rivers

Region Relative score

EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA EA LA SA SEAI SEAM SIDS SSA WANA

Source: Authors’ calculations

67.4 85.4 90.0 99.8 92.5 91.8 90.7 94.0 57.6 77.0 71.8 100.0 85.9 96.4 67.2 71.3 94.8 66.5 89.0 76.6 85.6 58.3 83.1 66.3 65.9 39.2 72.3 65.8 65.9 40.6 56.8 50.7 93.3 71.3 75.5 66.5 79.5 46.0 75.9 58.6

Rank

25 14 10 2 7 8 9 5 35 17 22 1 12 3 26 23 4 27 11 18 13 34 15 29 30 40 21 32 31 39 36 37 6 24 20 28 16 38 19 33

Rank deviation (BCR) -9 -25 -12 -29 -31 -32 -11 -13 24 -18 20 -12 -5 -7 22 11 -24 11 -15 -6 -17 15 -17 -8 25 33 20 24 22 25 30 34 -21 9 -5 5 -13 17 -14 -3

Priority Level

Medium High Very High Very High Very High Very High Very High Very High Low High Medium Very High High Very High Medium Medium Very High Medium High High High Low High Medium Medium Low Medium Low Low Low Low Low Very High Medium High Medium High Low High Low

35

Table 9: Unit indicators and ranking based on equity scores Region

System

EA

Pond Lake River Coast Coral Pond Lake River Coast Coral Pond Lake River Coast Coral Pond Lake River Coast Coral Pond Lake River Coast Coral Pond Lake River Coast Coral Pond Lake River Coast Coral Pond Lake River Coast Coral

LA

SA

SEAI

SEAM

SIDS

SSA

WANA

Relative score

Rank

Rank deviation (BCR)

57.7 57.7 61.3 54.1 45.0 55.6 41.7 41.7 50.0 41.7 69.0 80.5 80.5 80.5 72.4 59.1 59.1 59.1 60.0 46.1 64.1 70.8 70.8 68.8 50.6 58.6 58.6 58.6 69.6 62.3 76.1 100.0 100.0 95.7 87.0 52.6 63.9 63.9 60.1 52.6

28 29 19 31 37 30 38 39 35 40 13 5 6 7 9 22 23 24 21 36 15 10 11 14 34 25 26 27 12 18 8 1 2 3 4 32 16 17 20 33

23 1 -8 -3 26 23 22 24 -4 5 12 -21 -19 -15 7 14 -1 1 -10 23 6 -20 -18 -24 17 11 7 6 -28 8 2 -31 -31 -17 0 29 -21 -19 2 21

Source: Authors’ calculations

Priority Level

Medium Medium High Low Low Medium Low Low Low Low High Very High Very High Very High Very High Medium Medium Medium Medium Low High Very High High High Low Medium Medium Medium High High Very High Very High Very High Very High Very High Low High High High Low

36

Table 10: Final scores and rankings of units Rank Level 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

Very High Very High Very High Very High Very High Very High Very High Very High Very High Very High High High High High High High High High High High Medium Medium Medium Medium Medium Medium Medium Medium Medium Medium Low Low Low Low Low Low Low Low Low Low

Deviation from final rank Region System Final score Efficiency Environment Equity SA SA SSA SSA EA WANA SIDS SEAI SSA SEAI SEAM LA SSA SA SA SSA WANA WANA SEAM SA EA SEAM SEAI EA LA EA SEAM SEAM SIDS LA SEAI SIDS SEAI SIDS WANA EA WANA LA LA SIDS

Source: Author’s calculations

Ponds Coral Coral Ponds Ponds Ponds Coral Ponds Coasts Coral Ponds Ponds Lakes Coasts Lakes Rivers Coral Coasts Coral Rivers Rivers Lakes Coasts Lakes Rivers Coral Rivers Coasts Ponds Lakes Lakes Coasts Rivers Lakes Lakes Coasts Rivers Coasts Coral Rivers

85.3 78.9 73.3 64.9 63.0 61.8 59.8 59.6 55.8 55.1 53.7 53.5 51.5 51.0 50.1 49.7 49.7 49.1 46.9 46.8 46.3 46.1 46.1 45.7 45.3 45.2 44.6 44.6 42.1 42.1 41.7 40.3 39.2 38.6 37.3 35.6 35.4 34.9 34.8 34.7

0 0 0 0 0 0 0 0 -1 -1 1 1 -2 -1 -1 -2 0 0 0 -1 0 0 -1 0 1 1 0 -1 1 1 1 0 1 2 0 0 0 0 0 1

-2 -2 -2 -3 -2 -3 0 -3 0 0 -2 -2 0 1 0 0 -1 1 0 0 2 1 2 2 0 -1 1 2 -1 0 2 3 1 0 1 1 0 2 2 0

-1 0 0 0 -2 -3 -1 -2 0 -3 0 -1 1 1 1 1 -2 0 -2 1 1 2 0 0 -1 -1 1 1 0 -1 1 2 1 1 2 0 2 0 0 1

37

Figure 1: Schematic diagram of the priority setting process

Expert opinion

Modifiers

Shifters: Natural resource systems

Supply-Demand model

ECONOMIC SURPLUS ANALYSIS

Shifters: farming systems Cost of R & D investment

Related studies

Equity: subjective score, HDI Environment: subjective score Capacity-building: subjective score

Aggregate score (Modified Economic Surplus)

Research priorities