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), Sub-

Saharan 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, multi-

disciplinary 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 sub-

criteria 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

40
RMSD = (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:

1
MLDi = ∑ 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/e-

book/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 Units Objective Criteria Indicators Method
IRRI Regions Allocate budget Supply requirement Projected demand; yield gap Aggregated scoring with
Ecosystems shares by unit Equity Per capita calorie deprivation equal weights; modifiers
Sustainability Rice area under unfavorable include the gender
Presence of alternative environment (%) development index and
research supplier National agricultural research yield gap
spending
ICRISAT Themes Ordinal ranking; Efficiency NPV of production increase (with Benefit-cost analysis
Cumulative Equity modifiers) Aggregate scoring with
investment Internationality Poverty incidence and female equal weights
requirement Sustainability illiteracy rate
Simpson index
Subjective scoring
ILRI Themes Ordinal ranking; Efficiency NPV of production increase (with Benefit-cost analysis
Cumulative Equity modifiers) Aggregate scoring: 30%
investment Internationality Poverty index efficiency, 25% poverty,
requirement Sustainability Simpson index 20% sustainability, 15%
Capacity-building Subjective scoring capacity-building, 10%
Subjective scoring internationality

CIP Projects Ordinal ranking Efficiency NPV of production increase Benefit-cost analysis
Poverty Poverty index Aggregate scoring
CIMMYT Constraint x Ordinal ranking Efficiency Production index Aggregate scoring: 50%
region x Poverty Poverty incidence efficiency, 30% poverty,
ecology Presence of alternative Extent of subsistence farming 20% alternative supplier
research supplier

Sources: See references cited in the text.

 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 Countries
East Asia (EA) 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 Indonesia, Malaysia, Philippines

(SEA Island)
Southeast Asia Mainland Cambodia, Laos, Myanmar, Thailand, Vietnam
(SEA Mainland)
Small Island Developing Antigua and Barbuda, Bahamas, Barbados, Cape Verde, Comoros, Cook Islands, Cuba, Dominica, Dominican
States (SIDS) 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 Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Congo, Democratic
(SSA) 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 Afghanistan, Algeria, Bahrain, Egypt, Iran, Iraq, Jordan, Kuwait, Lebanon, Morocco, Qatar, Saudi Arabia, Syrian
Africa (WANA) 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 Marine capture Freshwater culture Brackishwater and TOTAL
marine culture (by region)

East Asia 2,293 14,813 16,979 16,641 50,725

Latin America 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
SIDS 19 512 17 33 581
Sub-Saharan Africa 1,719 3,413 51 53 5,236
WANA 429 2,885 386 841 4,540
TOTAL (by system) 8,181 54,463 23,396 26,385 112,425

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

 32

Table 5: Unit indicators and ranking based on economic efficiency

System Region Value of Annual Change in Benefit- Rank Priority
Output supply surplus Cost ratio Level
($ ‘000) shift (%) ($‘000)
Coasts EA 18,733 0.41 900 0.77 34 Low
LA 24,876 0.44 849 0.15 39 Low
SA 10,502 0.59 1,092 1.23 22 Medium
SEAI 12,901 0.32 1,133 0.90 31 Low
SEAM 19,226 0.42 1,157 0.63 38 Low
SIDS 0 0.42 0 0.00 40 Low
SSA 7,702 1.06 883 1.35 20 High
WANA 5,340 0.83 1,718 1.55 18 High
Coral EA 19 2.15 3 2.87 11 High
reefs LA 131 2.15 23 0.76 35 Low
SA 56 5.69 207 6.30 2 Very High
SEAI 2,431 2.15 819 2.74 13 High
SEAM 226 2.15 42 1.89 17 High
SIDS 2,242 2.15 379 2.97 10 Very High
SSA 747 6.44 361 5.11 4 Very High
WANA 405 2.49 260 2.76 12 High
Lakes EA 704 0.81 177 1.11 28 Medium
LA 923 0.97 91 2.21 16 High
SA 1,521 0.81 106 1.14 26 Medium
SEAI 672 1.13 36 1.15 24 Medium
SEAM 1,612 0.82 103 1.03 30 Medium
SIDS 30 0.58 2 1.38 19 High
SSA 1,680 0.94 79 0.85 32 Low
WANA 868 0.65 34 0.70 37 Low
Ponds EA 13,622 0.98 9,185 4.72 5 Very High
LA 2,419 0.98 1,473 4.38 7 Very High
SA 2,750 0.98 3,393 7.36 1 Very High
SEAI 2,467 0.98 1,462 4.17 8 Very High
SEAM 2,060 0.98 1,222 3.12 9 Very High
SIDS 168 0.98 164 2.55 14 High
SSA 144 0.98 118 4.66 6 Very High
WANA 1,048 0.98 918 5.29 3 Very High
Rivers EA 347 0.81 86 1.12 27 Medium
LA 455 0.97 51 2.51 15 High
SA 749 0.81 52 1.14 25 Medium
SEAI 331 1.13 18 1.15 23 Medium
SEAM 794 0.82 50 1.03 29 Medium
SIDS 15 0.58 1 1.25 21 Medium
SSA 1,680 0.94 79 0.85 33 Low
WANA 427 0.65 17 0.70 36 Low
Source: Authors’ calculations
 33

Table 6: Comparisons across priority rankings by economic efficiency indicator

Indicator Correlation MAD RMSD Correlation MAD RMSD
(ESA) (ESA) (ESA) (BCR) (BCR) (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

Relative to ESA
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
Relative to BCR
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

Source: Authors’ calculations

 34

Table 8: Unit indicators and ranking based on environment scores

System Region Relative score Rank Rank Priority Level
deviation
(BCR)
Coasts EA 67.4 25 -9 Medium
LA 85.4 14 -25 High
SA 90.0 10 -12 Very High
SEAI 99.8 2 -29 Very High
SEAM 92.5 7 -31 Very High
SIDS 91.8 8 -32 Very High
SSA 90.7 9 -11 Very High
WANA 94.0 5 -13 Very High
Coral reefs EA 57.6 35 24 Low
LA 77.0 17 -18 High
SA 71.8 22 20 Medium
SEAI 100.0 1 -12 Very High
SEAM 85.9 12 -5 High
SIDS 96.4 3 -7 Very High
SSA 67.2 26 22 Medium
WANA 71.3 23 11 Medium
Lakes EA 94.8 4 -24 Very High
LA 66.5 27 11 Medium
SA 89.0 11 -15 High
SEAI 76.6 18 -6 High
SEAM 85.6 13 -17 High
SIDS 58.3 34 15 Low
SSA 83.1 15 -17 High
WANA 66.3 29 -8 Medium
Ponds EA 65.9 30 25 Medium
LA 39.2 40 33 Low
SA 72.3 21 20 Medium
SEAI 65.8 32 24 Low
SEAM 65.9 31 22 Low
SIDS 40.6 39 25 Low
SSA 56.8 36 30 Low
WANA 50.7 37 34 Low
Rivers EA 93.3 6 -21 Very High
LA 71.3 24 9 Medium
SA 75.5 20 -5 High
SEAI 66.5 28 5 Medium
SEAM 79.5 16 -13 High
SIDS 46.0 38 17 Low
SSA 75.9 19 -14 High
WANA 58.6 33 -3 Low

Source: Authors’ calculations

 35

Table 9: Unit indicators and ranking based on equity scores

Region System Relative score Rank Rank deviation Priority Level
(BCR)
EA Pond 57.7 28 23 Medium
Lake 57.7 29 1 Medium
River 61.3 19 -8 High
Coast 54.1 31 -3 Low
Coral 45.0 37 26 Low
LA Pond 55.6 30 23 Medium
Lake 41.7 38 22 Low
River 41.7 39 24 Low
Coast 50.0 35 -4 Low
Coral 41.7 40 5 Low
SA Pond 69.0 13 12 High
Lake 80.5 5 -21 Very High
River 80.5 6 -19 Very High
Coast 80.5 7 -15 Very High
Coral 72.4 9 7 Very High
SEAI Pond 59.1 22 14 Medium
Lake 59.1 23 -1 Medium
River 59.1 24 1 Medium
Coast 60.0 21 -10 Medium
Coral 46.1 36 23 Low
SEAM Pond 64.1 15 6 High
Lake 70.8 10 -20 Very High
River 70.8 11 -18 High
Coast 68.8 14 -24 High
Coral 50.6 34 17 Low
SIDS Pond 58.6 25 11 Medium
Lake 58.6 26 7 Medium
River 58.6 27 6 Medium
Coast 69.6 12 -28 High
Coral 62.3 18 8 High
SSA Pond 76.1 8 2 Very High
Lake 100.0 1 -31 Very High
River 100.0 2 -31 Very High
Coast 95.7 3 -17 Very High
Coral 87.0 4 0 Very High
WANA Pond 52.6 32 29 Low
Lake 63.9 16 -21 High
River 63.9 17 -19 High
Coast 60.1 20 2 High
Coral 52.6 33 21 Low
Source: Authors’ calculations
 36

Table 10: Final scores and rankings of units

Deviation from final rank
Rank Level Region System Final score Efficiency Environment Equity
1 Very High SA Ponds 85.3 0 -2 -1
2 Very High SA Coral 78.9 0 -2 0
3 Very High SSA Coral 73.3 0 -2 0
4 Very High SSA Ponds 64.9 0 -3 0
5 Very High EA Ponds 63.0 0 -2 -2
6 Very High WANA Ponds 61.8 0 -3 -3
7 Very High SIDS Coral 59.8 0 0 -1
8 Very High SEAI Ponds 59.6 0 -3 -2
9 Very High SSA Coasts 55.8 -1 0 0
10 Very High SEAI Coral 55.1 -1 0 -3
11 High SEAM Ponds 53.7 1 -2 0
12 High LA Ponds 53.5 1 -2 -1
13 High SSA Lakes 51.5 -2 0 1
14 High SA Coasts 51.0 -1 1 1
15 High SA Lakes 50.1 -1 0 1
16 High SSA Rivers 49.7 -2 0 1
17 High WANA Coral 49.7 0 -1 -2
18 High WANA Coasts 49.1 0 1 0
19 High SEAM Coral 46.9 0 0 -2
20 High SA Rivers 46.8 -1 0 1
21 Medium EA Rivers 46.3 0 2 1
22 Medium SEAM Lakes 46.1 0 1 2
23 Medium SEAI Coasts 46.1 -1 2 0
24 Medium EA Lakes 45.7 0 2 0
25 Medium LA Rivers 45.3 1 0 -1
26 Medium EA Coral 45.2 1 -1 -1
27 Medium SEAM Rivers 44.6 0 1 1
28 Medium SEAM Coasts 44.6 -1 2 1
29 Medium SIDS Ponds 42.1 1 -1 0
30 Medium LA Lakes 42.1 1 0 -1
31 Low SEAI Lakes 41.7 1 2 1
32 Low SIDS Coasts 40.3 0 3 2
33 Low SEAI Rivers 39.2 1 1 1
34 Low SIDS Lakes 38.6 2 0 1
35 Low WANA Lakes 37.3 0 1 2
36 Low EA Coasts 35.6 0 1 0
37 Low WANA Rivers 35.4 0 0 2
38 Low LA Coasts 34.9 0 2 0
39 Low LA Coral 34.8 0 2 0
40 Low SIDS Rivers 34.7 1 0 1

Source: Author’s calculations

 37

Figure 1: Schematic diagram of the priority setting process

Expert opinion Modifiers

Equity: subjective score, HDI

Environment: subjective score
Shifters: Natural resource systems Capacity-building: subjective score

Supply-Demand Aggregate score

model
ECONOMIC Research
(Modified
SURPLUS Economic Surplus) priorities
ANALYSIS

Shifters: farming systems

Cost of R & D investment

Related studies