by Daniel D. Huppert
School of Marine Affairs, University of Washington
3707 Brooklyn Ave. NE, Seattle, WA 98105–6715
This paper reviews basic concepts of decision making under risk, and it describes risk assessment and risk management approaches developed for environmental protection decisions in the United States. Formal decision models quantify the value of strategies based upon probabilities of various outcome. Economic valuation of those outcomes can be used to rank strategies and to select the best ones. Fishery management decisions often can be assessed with this analytic method. However, differences between public perceptions of risk and technical measures of risk create problems. To reduce political opposition to implementing precautionary harvest policies, fishery managers could collaborate with interest groups (including the fishing industry) to communicate the risk information before adopting an arbitrary risk-avoidance strategy. The numerous types and sources of risk in fisheries are summarized in the paper, including those that affect the safety of fishermen, economic prosperity, and fish stock condition. Where fish stock collapse is the main risk, rational economic harvest policy safeguards against stock collapse by establishing an adaptive annual harvest quota and minimum stock level where harvest is curtailed. This is similar to biological benchmarks developed in the technical fisheries literature.
Depletion of the world's marine fish stocks by overfishing and habitat degradation, the most alarming risk facing fishery managers, accounts for current emphasis on the precautionary approach. Many people perceive that high rates of harvest can more-or-less permanently diminish economically valuable fish stocks and marine mammal populations. The collapses of Newfoundland cod, of Bering Sea red king crab, of Peruvian anchoveta, of north Atlantic herring, and of Antarctic blue whales are celebrated cases. If such collapses result from resource management decisions, then it is reasonable to expect managers to take more prudent actions to avoid such negative outcomes. Much of the technical literature addressing biological reference points (see Smith, Hunt, and Rivard 1993 or Hilborn and Peterman, this volume) concerns the appropriate quantification and presentation of the degree of fish stock risk inherent in harvest decisions. There are numerous other risks having important social and economic aspects. Examples include safety risks to seagoing workers, market price risks in the fishing industry, and risks of social disruption due to changes in fishery regulations, fishing industry technology, or resource abundance. Fisheries are largely economic in function, and fishery management has numerous economic and social, as well as fish stock conservation, objectives. This paper considers precautionary management in context of many sources of risks that are related to management decisions. I draw upon the risk assessment and decision analysis literature, and emphasize the social and economic dimensions of decisions regarding fish stock conservation. We begin by exploring concepts and processes common in risk assessment, extend the discussion to economics of risk management, and relate that information to the ongoing discussion of risk and biological reference points in the fisheries management literature.
Every human activity involves risk in the sense that negative and unintended outcomes occur in an unpredictable fashion. Traffic accidents, disease outbreaks, criminal activities are common sources of risk to individuals. Actions to avoid or control exposure to risk are equally common. Fire insurance on structures and personal liability insurance for auto accidents, for example, reduce the individual's risk of monetary loss. Individuals spend significant amounts to assure safety of financial assets, to avoid injury in accidents, and to insure personal safety from criminal activity. Still, individuals and groups intentionally and voluntarily take uninsured risks. Going to sea in fishing vessels poses risks of bodily harm, but promises economic benefits or increased incomes. Investing in fishing vessels or fish processing plants carries the risk of bankruptcy and poverty, but promises the careful investor a reasonable return on capital. People are generally aware of the trade-off between economic benefits and risks implicit in these actions. People everywhere routinely commit themselves to risky actions, but they also commit resources to reduce these risks. Precaution in fishery management is an extension of individual risk-taking behavior to broader community programs of fish conservation and economic development.
In fishery management, there are a number of risks whose significance in particular cases depends upon the objectives of the manager or fishery community. Risks include the stock collapses mentioned above, but also include more common occurrences such as temporary harvest reductions due to recruitment failures. A management strategy aimed at sustained yield has the risk of destablizing instead of stabilizing the harvest level. A plan of limited entry to a fishery may require the commitment of administrative resources while failing to diminish the fishing capacity and economic investment in the fishery. A management action intended to assist the economic development of a rural community may set into action social changes that cause disharmony and social distress. To think clearly about these risks it is useful to consider the general nature of risk and the specific sources of risk in fisheries.
There are two main sources of risk: Lack of Control and Lack of Information1. Lack of control implies that some events, like the roll of a dice, cannot be influenced significantly by people. Fishery managers do not control the weather, the ocean currents, the climate, or the mysterious processes of recruitment to fish stocks. Fishery managers are also not in control of social and economic processes that unfold as fishery plans and regulations are implemented. Even if we fully understand the physical and social processes involved, lack of control makes the outcomes uncertain. Lack of information causes risk independently of ability to control the underlying process. For example, a storm without warning may cause substantial loss of life at sea. Given appropriate weather information, we know in advance that a storm is imminent, fishing vessels will stay in port, and losses will be minimized. If we know that conditions for recruitment failure are occurring, we could reduce harvest levels in advance. But we do not have intimate knowledge of ocean conditions nor the understanding of the recruitment process necessary to make good predictions. Technical analysts can attempt to quantify the extent of risk associated with particular harvest levels, and this is a useful addition to the information set. But, harvest levels may still be too high or too low. Thus, where events are controlled, better information makes it possible to avoid some disasters and to use knowledge of a fishery system's operation to anticipate problems. Where information is good but control is lacking, it may be impossible to respond to imminent negative effects of change. Most fishery managers exercise fairly loose control over fishing activities, and most fishery information is rudimentary and unreliable. Hence, risks of unwanted negative outcomes are fairly high. Some common examples of fishery control and information problems in both the bio-physical and socio-economic spheres are listed in Table 1.
1 This discussion is a simplified version of that in MacCrimmon and Wehrung (1986)
|Components of Risk:|
|Determinants of Risk||Type and Size of Potential Loss||Chance of Potential Loss (Statistical risk)|
|1. Lack of Control|
|Oceanic Regime Change or Climate Change||Large potential losses and/or gains depending on what changes occur in fish stocks and yields. Disruption of existing investments and social organizations.||Unlikely over a short period; inevitable in long run.|
|Natural Disaster (Flood, Storm).||Acute losses of infrastructure in specific regions. Capital||Geographic distribution and frequency predictable from|
|Spread of Fish Disease or parasites||investments and lives lost. Lost production until disease cured or new stocks of fish established.||historical occurrence. Sporadic and poorly predictable.|
|Fishing Gear Improvements||May defeat effort-control measures, reduce economic||High likelihood of some effect.|
|Rapid Drop in Market Price||benefits of management. Substantial loss of fishing incomes||Can follow development of competing suppliers, e.g. aquaculture, or health risk.|
|Ecological effects of Introduced Species||Introduced species can cause decreased population of target fish stocks, reduced incomes.||Ecological linkages are very complex and difficult to anticipate. Low likelihood, but unpredictable.|
|2. Lack of Information|
|Poor forecasts of annual recruitment.||Short to medium term loss in harvest levels an incomes.||Fairly high chance of occurrence|
|Unknown “Threshold” for Stock Collapse||Collapse of stock and economic disaster for fishing community.||Relatively infrequent.|
|Poorly understood economics and social values of fishing||Non-optimal levels of fish stocks and fishing fleets, loss of economic benefits.||Prevalent in fisheries, high probability of occurrence.|
|Unknown costs to fleet of complying with management regulations.||Excessive costs of management, lower long term economic benefits of fishery.||Prevalent in fisheries managed by centralized agencies with low level efforts to coordinate with fishers.|
|Poor knowledge of non-use values of non-target species.||Failure to adequately conserve for amenity and other purposes. Loss of non-market economic values.||High chance, especially where fishery management regimes are controlled by fishing interests.|
“Risk” refers to situations where people able to formally calculate or intuitively gauge probabilities of losses based on past experience, experimentation, and/or statistical estimation. Repeated small-stakes gambling is a classic case of risk. It is possible to learn through experience the likelihood of a pay-off in games of dice, cards, and so forth. The gambler is accepting a risk of losing the initial bet in exchange for a probability of winning a much larger amount. Investing in a fishing vessel carries this sort of risk. Repeated experience shows that some vessels sink in accidents and storms; some fishing businesses go bankrupt due to poor luck or changing market conditions. The investor takes these risks into consideration. On the broader scale relevant to outcomes of fishery management, probabilistic risks can be calculated for recruitment levels, harvest rates, or economic returns to a fishing fleet. In many cases, the probabilities are gauged using standard statistical methods and based upon frequencies of observed past events.
Some people use a different term, “Uncertainty”, for situations where quantitative assessment of risk is impossible. “Uncertainty” would pertain to unpredictable events such as loss of fishing vessels in a war or in a tsunami, collapse of ecosystems due to invading organisms or climate change, collapse or sudden expansion of markets for the fishery's output due to technical innovations or medical research results. The conditions causing these outcomes can be understood and described, but lack of previous experience and consequent lack of information on likelihoods make statistical reasoning inapplicable. The importance of distinguishing risk from uncertainty has been emphasized, for example, in the economics of endangered species. In his version of the Safe Minimum Standard of conservation, for example, Bishop (1978) notes that we cannot know, even in a probabilistic sense, what large economic losses might be imposed on future generations by decisions to allow species extinction. He recommends taking actions to conserve all species until we can value the species and assess costs of preservation correctly.
In contrast to Bishop, Hirshleifer and Riley (1992, p. 9) consider the distinction between risk and uncertainty to be a sterile one. It does not matter whether risks can be quantified by statistical procedures. Hirshleifer and Riley deal solely with a “subjective” probability concept which they attribute to Savage (1954). Probability in this formulation is simply a degree of belief. They note that “Even in cases like the toss of a die where assigning “objective” probabilities appears possible, such an appearance is really illusory. That the chance of any single face turning up is one-sixth is a valid inference only if the die is a fair one -- a condition about which no one could ever be “objectively” certain. Decision-makers are therefore never in Knight's world of risk but instead always in his world of uncertainty.”
On a similar vein, in his major and influential work on ecological risk assessment Glenn Suter,III states that “frequentist concepts of risk are seldom applicable to ecological risk assessments because the endpoints are levels of effects on population or ecosystem properties, not the fate of their individual components” (p. 44). Suter claims that the most applicable type of probability is Bertrand Russell's (1948) notion of “credibility.” This notion is illustrated by reference to the weather forecaster, who uses a variety of models, information, and assumptions to estimate the probability of an unrepeated event. Cumulative probability curves can be developed based upon probabilistic models. The spread of the resulting probabilities depends upon “both the stochasticity of the environment and ignorance concerning measurable characteristics” of the system being forecasted. This concept is consistent with the Hirshleifer and Riley notion of uncertainty. I follow this logic in using the terms risk and uncertainty interchangeably to reflect subjective beliefs about the likelihood of outcomes. I assume that experience and scientific reasoning contribute to degrees of belief in various outcomes. But subjective judgements and other factors are also important in forming probabilities.
Some authors make the important distinction between “technical risk” and “perceived risk”2. Technical risk is probability-based assessment by experts using statistical methods, controlled experiments, and computer modeling. Technical risk uses the language of mathematics and expresses its conclusions in precise but often arcane terms. The community of people making technical assessments often have difficulty communicating their reasoning and conclusions to the non-technical audience. The practice of risk assessment described below is largely within the purview of technical risk analysis.
“Perceived risk” concerns the way in which the general public, those affected by fishery management decisions, understand the risks facing them and how they rank the various risks. Extensive research on the issue of public perceptions of risk finds that people often have difficulty in understanding probabilistic expressions of risk, even though they use similar concepts in assessing repeated risky decisions3. People frequently over-state risks of infrequent and relatively unknown events (e.g., nuclear reactor accidents) while under-stating risks of common, known events (e.g., auto accidents, disease due to smoking tobacco). Research also shows that people adjust their perceived risks in the face of new information. People may experience significant shifts in belief based upon single, disruptive events (e.g., an earthquake, major oil spill, disease outbreak, stock collapse,etc.). Further, perceived risk of an event is influenced by numerous contextual conditions. Among the factors listed by Merkhofer (1987;p.22) which have been found to influence the public's perception of risk are (a) severity of consequences, (b) familiarity of the risk, (c) reversibility of consequences, (d) impact on children, (e) whether distribution of consequences is equitable, (f) whether the risk is associated with dreaded fears (e.g., cancer risks), and (g) whether the risk is taken voluntarily or imposed.
The discrepancy between expert assessment and public perception of risk may be attributable in part to differences in information. Presumably, the scientists have data covering a wider range of empirical experience. If so, the public may need to be educated about the nature of the risks. But the fundamental differences in the way people commonly perceive risks are not necessarily susceptible to education. These and other sources of difference between expert and public risk perceptions cannot be taken casually. In particular, economic assessment of the costs of risk avoidance or benefits of risk-taking often rely on expressions of concern or willingness to pay derived from analysis of individual actions or responses. This is clearly true of economic analyses using market demand curves to assess economic values, and is likewise true of more modern methods of measuring non-market values, such as contingent valuation research. As a consequence, one cannot understand the degree of risk-avoidance or risk acceptance exercised by the public without direct investigation of the perceived risks. Public programs to avoid risks must include efforts to understand the perceived risks and to inform the public about the meaning of technical risk assessment. This two-way process of “risk communication” is normally the responsibility of public agencies and government officials.
2 I borrow this distinction from Leiss and Chociolko (1994, pp 36–37)
3 Kahneman, D., et al. (1982)
Adam Finkel defines risk assessment as “a multidisciplinary method … for estimating the probability and severity of hazards to human health, safety, and the natural environment.”4 A quantitative risk assessment provides:
Qualitative descriptions of the type and magnitude of adverse effect or hazard.
Numerical estimates of the probability of the hazard.
Discussions of the knowledge base on which the predictions of hazard are made. Risk assessment uses science to determine the probability of losses and to estimate the magnitude of the potential loss. The large and growing literature on risk assessment covers auto safety, engineering safety of offshore oil platforms, nuclear fuel disposal, pharmaceutical drug testing, environmental hazards from agricultural chemical applications, psychological risks associated with child abuse and suicide, investing in the stock market, water and air quality regulations, and global climate change. Any program or policy which significantly involves multiple outcomes with uncertainty can be subjected to risk assessment.
Comparative risk assessments are used to rank environmental risks and to determine which should be addressed first with limited resources. In the human health risk area, comparative risk assessments quantify the number of expected deaths per 100,000 population due to hazards posed, for example, by nuclear power plant accidents, radon contamination in private homes, excessive nitrates in drinking water, and airborne particulates from coal-fired electrical generating plants. The US Environmental Protection Agency has adopted risk-based priority setting methods that quantify the relative risks associated with various hazards5. The agency uses this information to decide whether a program to reduce levels of risk of one type should have priority over another. The goal of risk-based priority setting is to balance the environmental risks permitted against the cost of risk reduction and against competing risks6. In establishing budgetary priorities, the comparative risk assessment approach is linked to cost-effectiveness analysis. That is, given the budget available, it helps to select the mix of risk reduction actions that yield the greatest overall reduction in human health risks. When the kinds of risks being compared differ in significant ways(e.g., risk of morbidity versus immediate death, risk of birth defects versus risk of airline crashes), the comparative risk assessment must confront the problem of quantifying relative values, which raises conceptual complications often encountered in economic benefit-cost analysis. Without directly confronting this issue of relative value, comparative risk assessment can provide little guidance in setting priorities. Nevertheless, by identifying, describing, and quantifying what is risked and by whom, technical risk assessment can lead in a logical way to risk management.
Comparative risk assessment leads to comparisons of one risk with another (called “risk-risk” comparisons). For example, adopting groundfish trawls may risk disruption of benthic habitat and concentration of economic wealth, while use of gill nets may risk harvest of non-target migratory species and weak catch monitoring associated with widespread small-scale fishing fleets. Where benefits of risk-taking are calculable, a “risk-benefit” tradeoff analysis is possible. For example, one could array the likely economic benefits of increased harvest rates along with the associated risks of stock collapse. Similarly, one could assess the risk associated with introducing drift gill nets versus the economic benefits likely to result from improving gear efficiency. Where losses associated with the risks also can be estimated, one could perform a full “benefit-cost analysis”. In the benefit-cost framework, we subtract the expected economic losses from the expected economic gains for a particular action. Where net benefits are positive, the policy has potentially acceptable consequences. A major difficulty in using the net economic benefit criterion for making decisions is that the burden of losses and reward of benefits may be imposed on distinctly different people. Hence the question of equity in the burden of losses is an important policy issue in its own right.
4 Finkel and Golding (1994; p. 6)
5 See Kent and Allen's description of the EPA system in Chapter 4 of Finkel and Golding (1994)
6 Paraphrased from Glenn Suter, II (1993; p. 3)
Risk management goes beyond assessing the probability and size of possible losses in hazardous conditions, and beyond ranking actions by comparative risk. It entails deliberate attempts to reduce, diversify, or insure against risks. One means of reducing risk is to take cautious steps, planning ahead to avoid incurring excessive risks. Another is to eliminate hazardous activities entirely, but this is of limited use in the uncertain world of fishery management. Another is to spread risks more broadly and to diversify activities, making the aggregate program less subject to large negative outcomes. Explicit management of risks is, of course, not a technical analysis, but rather involves responsible agencies or community groups with authority to make decisions. Public decision processes are typically run by representatives seeking input from individuals, and this process is attentive to individual concerns. One cannot substitute expert opinion, based on technical risk assessment, for public perception on the assumption that a more fully educated public would agree with the experts. It apparently helps little for experts to “educate” citizens to the fact that nuclear power plants are safer that alternate sources of energy, for example. Similar dilemma's arise in marine fisheries. In the United States, for example, public pressure prevented the National Marine Fisheries Service from taking the gray whale off its endangered species list for many years after it was technically possible to do so. Hence, a direct and rigorous examination of what the public thinks the risks are, and how the public ranks various risks would be a useful adjunct to the model-based risk assessment methodology.
Leiss and Chociolko (1994) provide substantial evidence that intractable public controversies over management of health and environmental risks often stem from disagreements among technical experts on the magnitude and degree of risks. Further, these technical differences often stem from subjective judgements made in the risk assessment process. When technical disputes are combined with disparities between technical and publicly perceived risks, there is a tendency for the public to distrust the technical experts and for the technical experts to distrust the public statements by interests groups. Another important concern is that every decision to accept or control risk is paired with an assignment of responsibility for dealing with the consequences of taking risky actions. For example, the manufacturer of agricultural pesticides may demonstrate technically that there is little danger of environmental effects. But it is also likely that the manufacturer seeks to shift the responsibility for cleaning up waterways and compensating those affected by fish kills when the occasional accident occurs. This shifting of responsibility is another reason the public remains skeptical of public policies directed by technical risk assessment. One strong conclusion from Leiss and Chociolko is that risk management must involve a process of risk communication and of negotiation between parties with contesting interests. A negotiated agreement needs to define what risks are acceptable and who takes on the responsibility for those risks.
To clarify the nature of technical risk assessment in fisheries, it is helpful to use a framework from the economics of uncertainty and information theory (Hirshleifer and Riley, 1979 and 1992). First, there are alternative actions that can be taken. In a fishery we can think of these as various forms of regulations, institutional designs, incentive systems, and research strategies. It is obvious that the actions are uniquely human and are intended to achieve ends determined by humans. Actions may be single, once-for-all measures (also called “terminal” actions) such as prohibition of marine mammal harvests or limiting a fishery to hook and line gear. They may be repetitive actions such as annual setting of Total Allowable Catch. They may be contingency plans (e.g., a formula relating annual TAC to stock assessments). They may be informational actions or actively adaptive actions involving research programs or experimental fishing.
Second, there are states of the world (or simply States) which are largely uncontrollable and only partially observable, understandable, and predictable. The life history characteristics of fish, the mechanisms governing recruitment, the trophic dynamics among fish and their ecosystem co-habitants at various life stages, and the dynamics of ocean currents and primary productivity are important components of the State. For fishery managers there are social aspects to the State as well. The technology of fishing, economic characteristics of fishing fleet, the rituals and taboos that drive fishing behavior, and dynamics of fish marketing systems are examples of these social dimensions. Researchers seek constantly to improve knowledge of the State of the world, but its complexity eludes us7. We develop probability concepts to give shape to our ignorance and to quantify our inability to predict and to establish ranges of uncertainty.
Third, the State of the world determines how particular actions lead to specific outcomes or consequences. Outcomes are multidimensional, having both natural and social components. For example, adoption of a “conservative” fishing rate is generally thought to result in relatively small and benign changes in ecosystem structure, to long term sustainability or even stability of harvests, and to relatively low levels of aggregate economic production (food, recreation, subsistence, or whatever). An outcome of socially unrestricted fishing(e.g. open access) in the face of rising market prices and improving fishing technology is likely to be stock depletion, significant ecosystem restructuring, and unstable supplies of seafood. How much sustainability or depletion is caused by any specific action is, of course, uncertain. If the actions considered are complex, for example a harvest quota strategy involving adaptive management, the definition of the outcome is also complex. In economics as in ecology, everything is connected to everything else. There is seldom an unmitigated “good”. High levels of harvest are linked to lower prices and elevated harvest costs. That is good for consumers, but not so good for producers.
7 Jack Ward Thomas, Director of the US Forest Service, at the “Salmon Summit” in Portland Oregon in 1991, noted that “the ecosystem is not only more complex than we think, it is more complex than we can think.”
Fourth, the decision maker's preferences reflect individual preferences for various outcomes8. One may prefer high, unstable levels of fish production or low and more stable level of production. Regarding preferences, economics is typically non-judgmental; it takes people as they are (or they say they are), and accepts satisfaction of people's preferences as an adequate goal of decision making. Preferences may distinguish between any and all social and ecological attributes of the outcomes, and may even be expressed directly for States of the World. People may prefer the ocean with whales in it even if the presence of whales has no influence on the production of anything else of value, a situation giving rise to the economist's notion of existence (or passive use) value. A preference-scaling function is used to rank these values among the various outcomes. For simplicity, a complex economic outcome might be evaluated in terms of expected net present values of economic returns from the fishery. Preferences, however, may pertain also to the distribution of income or wealth among people -- an aspect that we broadly refer to as “economic equity”. Hence, use of whales for food diminishes the existence value held by mammal conservationists, while it increases the use value enjoyed by harvesters. Preference ranking by decision makers must deal with this inter-personal distributional issue9. Finally, the very presence of risk implies that a given action can result in various outcomes. Each outcome has a likelihood or probability. Hence, we need a preference ranking function that deals with multiple and uncertain outcomes.
Figure 1. A Framework for Risky Decisions
| ||State 1 |
Population Stable & Resilient
|State 2 |
Population Variable & Resilient
|State 3 |
Population Variable & Fragile
|Action 1 - Fixed Quota||v(C11)||v(C12)||v(C13)||U(A1)|
|Action 2 - Harvest Adaptively||v(C21)||v(C22)||v(C23)||U(A2)|
|with Ave. Catch near MSY|| || || || |
|Action 3 - Harvest Cautiously with low Ave. Annual Catch||v(C31)||v(C32)||v(C33)||U(A3)|
|Subjective Likelihood of State||p1||p2||p3|| |
8 I am using the abstraction of “a decision maker” in order to avoid the extensive extra language I would need to introduce the concepts of social or collective preferences. It is assumed that the decision maker somehow reflects collective preferences. Aggregating from individual preferences to a social or collective preference-scaling function requires consideration of inter-personal and inter-generational effects. In practice, the use of aggregate preference functions in economic research generally relies on some strong simplifying assumptions (e.g., that individuals are all affected in the same way by outcomes and have the same preferences). In more complex formulations, various social or economic classes may be identified and relative weights could be placed upon economic benefits for each class
9 A deep issue concerns the ethical force of preferences or values concerning others behavior. We accept the notion that our feelings about another's behavior, when that has only indirect or ephemeral affects on us, do not create a strong ethical obligation on the other person. This is at the heart of social toleration of aberrant grooming habits, religious practices, etc. That ethical tolerance apparently extends to the Hindu prohibition on harming cattle, for example. We do not accept an ethical duty on the part of non-Hindus to comply with that prohibition. Similar ethical tolerance could be accorded marine mammal harvests, bycatch, and other “demonized” aspects of fishing
We can illustrate these rather cryptic concepts with an example. With three possible states of the world and three management actions, we can depict the situation as in Figure 1 below. For concreteness we can think of State 1 as a world in which the fish population under management is “resilient”. Although it is affected by environmental variation and managers are unable to predict annual variations precisely or to measure the population with great accuracy, the range of variation in stock size is not great and it rebounds readily from episodes of excessive fishing. Consequently, it is relatively easy for managers to adjust fishing rates to rebuild stock abundance in response to declines. In State 2, the population variation covers a wide range (e.g., something like a coastal pelagic schooling fish stock) but the stock is still resilient in the sense that it will rebound over a reasonable period of time (5 – 10 years?). In State 3 the fish population is part of a complex, nonlinear ecosystem in which significant fishing triggers a change in trophic dynamics rendering the population prone to replacement by competing species. Once the population is significantly disturbed it declines to levels of near commercial extinction. Each combination of management action and State is identified with a particular consequence, C, and we presume the decision maker to have a utility function, v, which expresses a ranking of preferences across consequences.
A manager can select Action 1 -- an aggressive, MSY-type management strategy, Action 2 -- a strongly monitored and regulated adaptive management strategy, or Action 3 -- a minimal harvest rate strategy. The choice depends on which yields the greatest pay-off or value. For purposes of the example, we assume that v(C11)>v(C21)>v(C31), v(C12) < v(C22) > v(C32), and v(C13) < v(C23) < v(C33). Hence, a manager who is able to determine the state of the world chooses Action 1 in State 1, Action 2 in State 2, and Action 3 in State 3. It is easy to conclude that, without risk, the selection of management action is immediately connected to the State of the world and the consequences. This is the sort of decision consumers and producers are normally expected to make in microeconomic theory. If the state of the world is that your shoes are worn out; you can buy a new pair of shoes, and the outcome is obvious and immediate. However, once we introduce uncertainty, choosing an action is not equivalent to choosing anoutcome. For any given action, numerous consequences are possible. The shoes may turn out to hurt your feet, or a defect in manufacture may cause the sole to detach during the next rainstorm. The action -- buying shoes -- does not provide with certainty the desired outcome. Hence, rational behavior under uncertainty cannot be equivalent to choosing the outcome which maximizes the preference ranking function defined over outcomes. We must choose actions based on an evaluation of numerous possible outcomes and their likelihood.
In the simple example of Figure 1, the manager faces a risky decision because he cannot tell exactly which state of the world prevails. To resolve this, she may choose the action which maximizes the expected utility function defined as:
U(Ai) = p1v(C11) + p2v(C12) + p3v(C13). Here, the function U(Ai), is a von Neumann and Morgenstern utility function which is defined over actions having uncertain outcomes. This says that the utility of an action can be computed from the elementary utilities of consequences via a simple linear weighting, using probabilities (pi) as weights; i.e. the mathematical expectation or probability-weighted average. As stated by Hirshleifer and Riley: “The great contribution of Neumann and Morgenstern was to show that, given plausible assumptions about individual preferences, it is possible to construct a v(c) -- “cardinal” in that only positive linear transformations thereof are permissible -- function whose joint use with the Expected utility Rule will lead to the correct ordering of actions.” (p. 15). The cardinal v(c) function can be constructed in a special way called the reference lottery technique10.
For simplicity, we could assume that the consequences are quantified in terms of net increments to economic income, and that the elementary utility function v(c) assigns utility to income. Further, if the v(c) is a simple linear function like v(c) = ac, more income is preferred to less, and proportionally more income generates proportionally more utility. Here, the decision maker using the expected utility rule will seek to maximize expected income. In this construction the decision maker would be indifferent between any two prospects having the same expected income even the two prospects have difference risks of loss - he/she would have a risk neutral preference ordering. By extension, a decision process seeking to maximize a more complex concept of expected economic benefits (e.g., consumer and producer surplus) is acting in a risk neutral manner. This does not mean that the decision maker is careless or lacks prudence, it is simply a technical statement about the shape of the utility function. If the utility function is convex, like v(c) = alog(c), the decision maker will prefer the prospect with lower risk of failure. The resulting behavior is termed risk averse11. A risk averse person is willing to accept lower expected value in exchange for lower risk. A risk-preferring (also called risk-prone) decision maker would do the opposite -- choose a riskier option even if it had lower expected value. Generally speaking, most decision makers exhibit characteristics of risk aversion; although, when risks can be broadly diversified, individual decisions may appear to be risk neutral. This could be the case, for example, of small program decisions by a large government agency. Consistent risk-preferring behavior, when extended beyond moderate small-stakes gambling, is generally treated as a pathology, and often leads to ruin.
Elaboration of this formal model of decision making helps to focus attention on the risk assessment and evaluation on the following decisions components.
Outcomes or consequences of harvest strategies. Socio-economic dimensions of the outcomes may differ those typically examined in biological/ecological models. Economists concern themselves with net economic returns (or rents) generated by the fishery. They also develop more complicated models which deal with consumer surplus, willingness to pay (or sell) for non-market use values (e.g., recreation), existence or passive use values, short-run versus long-run costs and returns, employment and regional income generated by fisheries, and costs of management actions. In a simple fishery market model, for example, some outcomes of a harvest policy would include the mean and variance of market price, supply, incomes, and employment.
Preference-ranking functions over outcomes. Rigorous assessment and quantification of a publicly-acceptable preference ranking is critical, but is generally beyond the scope of fishery models. Simple preference rankings often used in fisheries research tend to be linear or logarithmic functions of average harvest or weighted combinations of average harvest and stock size. Economic preference rankings would incorporate the consumer benefits, producer benefits, non-market benefits, and all costs into a single scalar “net social benefits” measurement. An alternative is to display alternative ranking using different systems of weights for the component net benefits variables. If the fishery strategy is dynamic, the preference ranking function must deal with outcomes and preferences at different points in time. This raises the question of how to incorporate opportunity costs of capital and social time preference rates as discount factors.
10 This technique is explained in detail by Hirshleifer and Riley (p. 16–21)
11 Actually, this statement is not quite correct. Hirshleifer and Riley carefully show that the shape of the v(c) function depends upon both attitudes towards risk and marginal utility of income. A risk neutral decision maker with a declining marginal utility of income will have a convex v(c) - resulting in decisions that are commonly characterized as “risk averse”. I ignore this important distinction throughout, because the common practice in the fisheries literature is to label as “risk averse” any utility function that is convex in consequences
The harvest strategies to be considered. Biologists' models frequently examine fixed catch or fixed fishing mortality rate or fixed escapement strategies. There is no particular reason to choose these options except that they are easily represented algebraically, given the usual structure of the population models. Economic optimization models typically yield feed-back control strategies which assign annual quotas based upon perceptions of the state of the fishery. In simple cases, this control policy sets a threshold stock size below which no harvest is allowed, because harvesting is less valuable than investment in larger stock size. Above the threshold stock, annual harvest level is an increasing and generally nonlinear function of measured stock size.
Assessment of risks. Assigning probabilities to the alternative outcomes under each state of the World involves both a decision regarding which States to consider and how to quantify the risks. Introduction of measurement errors, model errors, noise and implementation errors into management models has been well developed in biological models. Economic models are not as tractable, since it is nearly impossible to conceive of risks as one dimensional. A risk of lower harvest is tied to a risk of higher price. The two variables affect consumers and producers differently. And the outcomes in one fishery tend to affect conditions in other fisheries.
Much of the hard scientific research effort in fisheries decision analysis emphasizes alternative harvest management strategies, quantifies consequences in terms of average harvest and variance of harvests over time, and assesses the probabilities of those consequences. Social and economic consequences, such as income and employment in the fishery, can be evaluated as well, and these are often as difficult to predict and measure as biological and ecological consequences. For example, commentators during the Symposium on Fishery Management: Global Trends in June 1994 noted the difficulty of forecasting the effect of individual transferable quotas (ITQs) on the economic structure of the fishing industry12. Experience shows that in some fisheries ITQs encourage consolidation and increased scale of firms, while in other fisheries they strengthen the small, owner-operated fishing firms. Further, because neither economic benefits nor aquatic ecosystems are directly observable, measurement of ecosystem status and measurement of economic benefits are equally difficult. An additional problem is that the economic values being measured are subject to exogenous change over time. From the perspective of a given fishery management regime, a change in foreign exchange rate or the emergence of a new product which competes with the managed fish species will cause an unexpected shift in the economic value generated by the fishery.
Regarding non-market values, special skills and research efforts are needed to adequately quantify economic benefits13. Recreational values are important for some fish species in the US and Europe, yet the economic value of recreation is not well documented in the statistics normally collected by management agencies. Studies to document the value of recreational fishing using well known techniques are expensive and therefore scarce. Further, the concepts involved in extending economic value to non-market goods are misunderstood by many authorities making management decision. These problems are even worse in the area of existence value or passive use value, which can be estimated only by well-designed and executed contingent valuation surveys. In conducting such surveys for measuring, the research approach must take into account the behavior of people in responding to solicitations. The research methods are fairly well developed now14, but there are significant outstanding controversies concerning the accuracy and usefulness of contingent valuation methods for estimating benefits, for determining adequate compensation, and for ranking of social program objectives.
12 The Proceedings of that symposium are still in preparation. The particular comments are attributed to Dr. James Wilen, Dr. Harlan Lampe, and Philip Major who were drawing upon experience in Canada, Chile, and New Zealand respectively
13 Freeman (1993) provides a useful overview of the entire field of non-market values
One logical response to inaccuracy and in-comparability of economic value estimates would be to stop short of providing a single, comprehensive measure of economic benefits (or expected utility). Instead, one could provide the decision maker 'with a display of alternative social and economic consequences for each action under consideration. The display would logically include some information about variances and covariances as well as means of the key variables. This approach raises the problem that alternative actions and consequences are infinite in number. So a process of reducing the set of options to those most appropriate and acceptable needs to be pursued along with the technical analysis. For example, in presenting technical information to the Fishery Management Councils in the US, technical advisors often provide a range of estimates for alternative management approaches. These may include the average and statistical range of harvests or stock abundance under exploitation strategies; and this is frequently augmented with expected distribution of harvests among important gear groups and/or fishing ports. Economic value of harvest, recreational share of harvest, and economic impacts of harvesting activity on coastal communities are also provided occasionally to Fishery Management Councils. This approach is consistent with multi-objective framework for decision making and seems to be the approach increasingly advocated by researchers involved in providing information to managers15.
Models of fishery management that incorporate uncertainty in population dynamics, in measurement of stocks, and in behavior of fishing effort in response to management are well developed. These seem to be generally consistent with the framework described in Figure 1. The compendium of papers presented at a workshop in Halifax, Nova Scotia edited by Stephen J. Smith, Joseph J. Hunt, and Denis Rivard (1993) is a recent example of that literature. To date much effort is focused on foreseeing the possible biological outcomes of various harvest strategies and estimating the probabilities of various fish stock levels, recruitment failures, and stock collapse. Little analytical effort seems to be focused on other kinds of risk, and little is devoted to deep understanding of the social and economic consequences of the outcomes (e.g., the utility functions).
It is clear that formal fishery decision models constitute a form of ecological risk assessment. The National Research Council's Committee on Risk Assessment used a study of George's Bank fishery as an example of risk assessment, and it was their only example of a well quantified ecological risk assessment16. Fisheries research is far ahead of other branches of ecological risk assessment, at least is developing quantitative models. The frameworks used in most fisheries models pay special attention to the process of quantifying particular outcomes using specific assumptions regarding fishing strategies. These strategies are frequently expressed in terms of a fishing mortality rate (F) or a harvest amount (C) or a combination of F and minimum spawning stock biomass (SSB). Recent innovations in the techniques include bootstrapping and Monte Carlo modeling of fisheries populations under exploitation. The sources of uncertainty introduced in such models focus on biological/ecological concepts and measurement error problems. For example, recruitment is often taken as a random process -- either uniformly distributed or normally distributed about some stock-recruitment function. Error in the estimation of stock-recruitment functions and randomness in natural processes are recognized as major sources of risk. In addition, errors in measurement of stock size, natural mortality, fishing effort rate can be introduced to such models.
14 See Mitchell and Carson (1989)
15 Hilborn, Pikitch and Francis (1993)
Technical analysts typically search for strategies that are robust to the underlying modeling errors, measurement errors, and unpredictable fluctuations in recruitment. Examples of such strategies are the familiar F0.1 fishing rate, establishment of threshold stock size, and constant escapement strategies. Biological reference points are extensively reviewed in the Smith, Hunt, and Rivard report on the Halifax Workshop, and the consequences of adopting various such reference points in several fisheries are examined there.
Given the discussion of risk assessment and risk management objectives from above, it is clear that the models pursue technical risk assessment and that they are clearly intend to feed directly into risk management processes. Because most such models focus rather narrowly on one or two simple outcomes -- catches (average or sustained) and stock biomass, they provide limited information for managers concerned about social and other aspects of management. The Halifax Workshop report highlights the important distinction between management objectives and risk assessment. For example, the Working Group Reports (pp. 5–12) concludes that additional work is needed to develop an analysis of outcomes in the social and economic dimensions and that the process of management needs to formally incorporate these diverse analyses along with input from client groups. “Biological objectives” are conceived as ways of limiting the scope of harvest strategies considered by managers. Working Group #1 concluded that “Science should not presume or establish objectives or socioeconomic strategies”, but it should determine biological constraints on the management system and translate the scientists level of uncertainty into ranges of possible outcomes. On the other hand, some papers in the Workshop proceedings17 suggest particular conservation strategies without explicit reference to any of the social or economic effects.
From the perspective of social science, fishery decision analyses available in scientific journals illustrate the kinds of risks posed by uncertainty in the biology and stock assessment side of the business. Less has been learned of the social and economic aspects, but there are certainly a number of useful studies and conclusions worthy of consideration. One approach is to expand a biological model by including economic benefits and costs in the objective function. Some examples of this are published in the Halifax Workshop report. Another example that I am particularly familiar with is the analysis of alternative harvest strategies for the northern anchovy fishery off southern California18. Because the anchovy stock was considered of major importance as a prey species for more valuable sport-caught fish, a key prey species for the brown pelican nesting colonies, and a prime live-bait source for recreational fishermen, there was substantial interest in being cautious about the exploitation strategy. One of the analytical efforts was a dynamic programming model which focused on optimal net economic return from the commercial anchovy fishery. The economic criterion was the expected discounted commercial value minus cost of harvesting over time. Because the anchovy stock tended to vary significantly from year to year, we introduced a stochastic error term to the estimated year-to-year population biomass transition model. The mathematical statement of the optimization exercise was expressed as:
16 National Research Council (1993) p.252
17 G. Thompson. (1993)
Bt = G(Bt, yt)et
The variables are defined as follows: h is the annual yield, B is the anchovy biomass, i is the annual discount rate, and e is a log-normal random error. V(..) is a net economic value function defined to include the economic value contributed by the combined fishing/processing industry. G(..) is the estimated year-to-year biomass transition equation. The economic model of the harvesting assumed fishing costs were simply proportional to number of days fished, but also recognized that catch per day of fishing was non-linearly related to biomass. This resulted in the following cost per unit catch equation:
(2) c = aB - .04
This implies that cost per unit harvest increases as biomass declines. Given the fixed demand curve for the product of the fishery (fish meal), and this cost function, it is relatively straightforward to see that there is some minimum biomass at which the annual net value of harvest will be zero. This level of biomass could be considered an “economic reference point” of sorts. When biomass is below this level, it is sensible to have a harvest rate of zero -- to allow the stock to grow at its natural rate. We solved the problem by discrete dynamic programming, and obtained an optimal harvest strategy which assigns a specific harvest level to each year's spawning biomass estimate. Because the optimization problem is dynamic, the optimum harvest strategy must satisfy the following condition: at each perceived stock size (i.e., State of the World) the harvest should be adjusted until the marginal current value (contribution to net economic value in the current year) just equals the marginal value, discounted one year, of increased biomass. An increase in biomass generates a future value because it decreases future harvesting costs and because in contributes to future sustainable yield. As noted by Clark and Munro (1975), the dynamic economic optimum treats biomass as an investment. We apply the standard investment criterion of adding more to the asset so long as the return on investment is at least equal to opportunity costs of investment. The opportunity cost is represented by the discount rate times the current net value of a unit harvested.
18 The models are documented in Huppert (1981) and Huppert, MacCall, and Stauffer (1980)
In the dynamic formulation of the harvest optimizing problem, the economic reference point is a biomass greater than the level at which annual net value equals zero. Not only do we want a biomass large enough to yield positive net benefits (market value of fish less costs of fishing), but we also want to cover the opportunity costs of discounted net revenues generated by larger biomass. Hence, the economic “reference point” is sensitive to the discount rate, but will generally be a biomass at least equal to the level where cost per unit just equals price per unit. It is well known that competitive market economies with no property rights in fish stocks commonly overexploit fisheries, depleting stocks well below this economic reference point19. As further explained by Smith (1968) and Clark (1973) there may even be pathological cases in which a present value maximizing strategy drives the stock to very low levels, possibly even to extinction. In those cases, the managers need to explore public perceptions and values to determine whether extinction of the stock would generate significant non-market losses not incorporated in the net present value formula.
In some respects, the economically optimal harvest policy developed for the anchovy stock, and similar harvest strategies developed elsewhere, have characteristics of a precautionary policy. That is, the economically sensible thing is to protect the stock when spawning biomass is low. A complete shut-down of the commercial fishery at low stock biomass is completely consistent with the usual notion of economic rationality. This is true even though future returns are discounted, even though the decision criterion is risk-neutral, and even though non-market or non-economic ethical values were not considered. Whether proponents of the precautionary approach accept this economic optimization approach to biomass management will depend in large part on whether it adequately represents the perceived risks and is adequately protective of fish stocks.
That this conclusion is not widely understood or appreciated may perhaps be attributed to two circumstances. First, many technically competent fisheries researchers (and certainly many influential conservation leaders) have little or no serious training in quantitative economics. Second, many people have come to understand “economics” to be whatever local, commercial interests say it is. Entrenched economic interest groups often have very narrow agendas and myopic views of conservation. This is often due to a history of operating in an open access fishery, where individual incentives to conserve fish are very weak at best. It may also be due to the notion of “individual risk” that I mentioned earlier. The problem is that an individual fishing operator may be unable to survive an extended fishery closure because she is financing capital investment in vessels through bank loans. In this case, even though the individual understands that a period of stock re-building is the best strategy for the fishery, she may feel compelled to advocate continued fishing, hoping that the stock will recover anyway. Hence, individual survival incentives may drive industry economic interests to fight for collectively irrational policies. Another way to state this is that the burden for risk-reduction may fall disproportinately on the small fishing operations. The solution to this dilemma is not be to criticize economics or the use of economic criteria in establishing precautionary fishery management approaches, but rather to alter the social system so that individuals can survive the occasional closed fishery, that is to reduce the individual risk.
A component of a fully developed precautionary fishing policy could be a form of insurance or compensation fund that could be triggered when stocks decline below economic reference levels. A credible commitment by the authorities to help individual firms and families to survive these situations might go a long way towards bringing rational attitudes to the management decision process. There is, of course, another side to this policy dilemma. The establishment of such a policy could become a continual subsidy to the fishing community, and this might encourage fishing fleets to expand because the compensation will essentially lower long run average private costs of fishing. A different approach would be to create and enforce individual, durable fishing rights in the fishery individual fishing quotas (IFQs) or individual tradeable quotas (ITQs). A successful conservation program will generate capital asset values for the ITQ holders. Financially strapped individual fishers could draw upon that asset to survive occasional low quotas or prices. Banks are more willing to extend credit to owners of these assets. This may alleviate individual risks sufficiently to moderate short run anti-conservation incentives.
19 See any textbook of fisheries economics or bioeconomics, for example L. Anderson (1986) or C. Clark (1976)
When species extinction is threatened, a new set of social and economic considerations becomes relevant. Economists often introduce the notions of “existence value” or “passive use value” to represent the value that people place on maintaining a natural asset even if those people expect never to consume or otherwise directly use the asset. Existence values have been estimated for everything from clean air in Los Angeles, to marine mammals, to grizzly bears in Montana. The technique used to estimate these values is called “contingent valuation” because it typically calls upon survey respondents to make decisions regarding payment, contingent on apostulated market. Recently, John Loomis completed a survey from which he estimated a value in the US for elimination of two hydroelectric dams on the Elwha river in Washington state. The policy issue is restoration of salmon runs to a part of the river that has been inaccessible to salmon for 50 years. His estimated total annual value for the US population (expressed by the survey respondents as willingness to pay higher taxes to finance the dam removal) ranged from roughly $3 to 6 billion. The large value is likely due to salmon's status as a “charismatic megafauna”, at least in the Pacific Northwest region of the US. It is unlikely that a north sea plaice would generate a similar existence value, but you never know what the value might be until you take the effort to estimate it. The existence value of a fish stock would be expected to become a greater and greater portion of the total economic value as the stock is pushed towards extinction. Hence, when included in the economic assessment of harvest strategy, these existence values militate against depleting stocks to a point that raises the likelihood of extinction.
The greatest threats of extinction in capture fisheries occur when the stocks affected have low production rates or depend upon specific ecological niches that are altered by fishing. Significant levels of harvest can drive stocks to below threshold levels before managers have an opportunity to evaluate the risk and arrest the decline. Here, the precautionary approach would call for the slow build-up of the fishery while information collection and analysis accumulates. Most species with low production have predictable characteristics (sharks and mammals have low reproduction rates, rockfishes and orange roughly live to advanced ages) that can be assessed in advance. Sensitivity to ecosystem structure is essentially more difficult to appraise. For example, some mammalogists suspect that harvesting at a moderate levels in the Bering sea sets in motion a long term process of ecosystem restructuring that threatens the northern sea lion. Managers expect the evidence of imminent extinction to increase over time, giving the decision makers time to adapt the associated harvesting activity while monitoring the sea lion population. Whether this approach to species conservation is successful will hinge on uncertainties concerning the ecological linkages between species and errors in measurement of population sizes and trajectories. In the US, where people place a lot of passive use value on sea lions, there is pressure to reduce and restructure the fishery more than fishery managers suggest. An economic approach to the decision would call for evaluation of the passive use values along with a net benefit analysis of alternative harvest rates and fishing patterns. Whether more extreme restrictions on fishing are appropriate would depend upon whether the costs of precautionary measures exceed the benefits decreased chance of extinction.
In his economic analysis of species extinction Bishop (1978, 1993) relies on the twin notions of irreversibility and uncertainty to justify a more cautious decision criteria which he calls the Safe Minimum Standard approach. Because the full consequences of species loss is unknown and extinction is irreversible, Bishop reasons that the future economic loss due to extinction could be very large and not calculable. Further, lacking probability estimates associated with the unknown losses, we cannot maximize expected benefits. Bishop concludes that a reasonable economic decision criteria for endangered species is minimizing the maximum possible loss (minimax loss). If the potential long term losses are seen as arbitrarily large, we should always preserve the species. Bishop adds a proviso to his “safe minimum standard” approach that extinction may be selected if the social costs of preservation are intolerably high. In responding to the critique of Smith and Krutilla (1979), who suggested that a better approach would be to maximize the expected value of future net benefits, Bishop (1979) noted that he had in mind a situation of “true uncertainty”, where expected losses cannot be computed and entered into a net benefit assessment. This Safe Minimum Standard approach to endangered species has much in common with the so-called “precautionary principle”. Critics of that approach point out the excessive emphasis it places upon unmeasured large losses having very low probabilities -- a criticism that essentially says the decision criterion is too risk averse20.
A problem ignored by Bishop and his critics is that actions needed to save a species are also risky. It is uncertain that species conservation efforts will be successful. In terms of the formal decision model, this is another form of uncertainty about the State of the World. Montgomery and Brown (1992) develop a constructive analysis of this uncertainty in the context of the endangered spotted owl in the US Pacific Northwest forests. In their formulation, the probability of survival is the key variable. There is essentially nothing we can do to guarantee the long term survival of the species. Still, the larger the amount of forest that is preserved as owl habitat, the higher the probability of survival. They used two different models to generate the probabilities of owl survival one using a group of experts and the other an explicit population model. Directly linked to the magnitude of forest preservation is the opportunity cost of foregone timber production. This approach displays the cumulative economic cost and cumulative gains in survival probability associated with forest protection. The economic value of saving owls has also been estimated, but these numbers seem to be less reliable at present. The display of trade-offs between costs and owl survival is a useful example of cost-effectiveness analysis.
Finally, Walters (1986), Kai Lee (1994), and others emphasize that estimates of probabilities and of costs can be improved over time through adaptive management -- that is, by treating management actions as experiments from which we learn about the system being managed. The Safe Minimum Standard approach could be a reasonable risk management strategy when used as a first step in an adaptive learning strategy. That is, we would preserve every species even at high cost until we understand more about its role in the ecosystem, its chemical composition, etc. After that, the life-or-death decision could be based upon constructive estimation of survival probabilities and associated costs and benefits. To the extent that stock collapses and ecosystem catastrophes are analogs of a species extinction, the same approach could be used -- that is, take extreme caution and forego the short-term economic benefits of aggressive harvesting in order to avoid extreme outcomes. Devote resources to assessing the productivity of the fish stocks and to economic benefits of further expansion, and experiment with harvest levels of learn from experience. This is a narrow version of the precautionary approach that ignores social and economic consequences of reduced food supplies. However, most fishery decisions, even in risky environments, do not involve significant probabilities of catastrophic outcomes. Learning to scale the degree of caution to the likelihood of negative or catastrophic outcomes is the major task of risk assessment and management.
20 Bishop could respond, of course, that the critics can not logically assess how risky the strategy is since neither probabilities nor potential losses are quantifiable
Some people place such high negative value on the stock collapse or species extinction outcome that they act in accord with the “precautionary principle” -- permit no harvest until it can be proven safe. That commercial fishing interests view this approach as irrational stems from disagreement over values, not over technical risk assessment. No amount of scientific, economic, or social research will solve the problem of divergent values. But we can attempt to avoid allowing the fundamental differences in values becoming tangled up with differences in risk management approaches. This is where the risk communication and policy negotiation processes become extremely important.
Risks commonly affecting fisheries include market price fluctuations, operating cost increases, adverse weather patterns, onboard crew safety, and fishery-unrelated shifts in species composition of harvested stocks. Any significant shift in conditions that underlie economic returns to fishing pose a risk to those investing capital in or dedicating their lives to fishing. For example, reduced ex-vessel prices severely depressed incomes in the Alaska salmon fishing fleet during the record levels of salmon production in 1989–1992, and a similar drop in worldwide cod and related species prices reduced incomes in fisheries from Iceland to Alaska. The salmon industry experience suggests that world demand for salmon is elastic, and it has brought into question economic benefits of the vast investment in salmon hatcheries in Alaska. Managers of other large commercial fisheries may need to anticipate effects of harvest volume on market price, especially in fisheries (e.g., Norwegian farmed salmon, Alaska wild salmon, north Pacific walleye pollock, Peruvian anchovy) which contribute a large portion of the world's supply. If fishing fleet prosperity is a management objective, existence of a price-elastic demand suggests that managers should place less weight on large harvests in their objective functions.
Serious fluctuations in the costs of inputs to fishing are less frequent and should be less of a concern in developing management strategy. However, during the Arab oil embargoes of the 1970s, fishing fleets in the US were suddenly unable to buy or, if they could buy, unable to afford the normal quantity of fuel. There is little that fishery managers can do or are expected to do about input price risks. They can provide information to fishing firms that would help them choose fishing patterns that reduce exposure to input price risk. For example, if small, inshore fishing operations are less dependent on purchased inputs (fuel, manufactured gear) than larger offshore operations, then the type of fishing fleet developed could affect risk due to input price fluctuations.
Safety risks are a major factor in many US north Pacific commercial fisheries. An American Journal of Public Health paper21claims the death rate of 414 per 100,000 fishermen per year is 53 times the US national average industrial mortality rate. Especially in small boat fisheries operating under strict fishing season regimes, competitive open access fishing can encourage vessel operators to fish during high wind and wave conditions. The Seattle Times calls for “Congress and the North Pacific Fisheries Council to confront an outdated open-entry management that encourages people to risk their lives for somebody's seafood dinner. A system of individual quotas would change those incentives, and minimize the hazards.” That editorial perspective is shared by many participants and managers in the north Pacific. They claim the safety advantages of individual fishing quotas arise because each fishing vessel operator can choose to avoid bad weather without sacrificing ability to harvest a fair share of the annual quota. IFQs can reduce (but certainly not eliminate) the economic incentive to take undue risks at sea.
21 Quoted in a Seattle Times Editorial (Tuesday, March 1, 1994)
Finally, there is increasing evidence that ocean “regime shifts” cause significant and widespread changes in abundance and species composition of ocean fish stocks. For example, Hare and Francis22 find a shift in the Aleutian low is associated with changes in north Pacific salmon abundance. If it is true that these kinds of shifts have unpredictable and uncontrollable effects on fish stock abundance, then the associated risk is not avoidable. Even so, the presence of this sort of risk has an important implication for fishery managers. Since the maintenance of stock biomass as an economic investment -- which increases future potential harvest levels and decreases average cost of harvest -- ecological risks translate into economic risk. The expected return on that investment is less when random stock collapse may occur. Managers could respond by investing less in the stock, harvesting more and maintaining a smaller fish stock over time. The presence of this “ecological risk” alters the trade-off between fairly certain short term economic benefits of higher current harvests and the less certain long term benefit of fish stock investment. One suspects that many “environmentalists” would take the opposite tack; they would reduce harvest levels in the face of ecological risk. Another response to unavoidable ecological risk would be to encourage development of multi-purpose fishing fleets, broadening the species and stocks supporting the economic enterprises, and hence reducing the exposure to risk of economic collapse from shifts in any particular stock.
When random stock collapse is considered to be a continuing risk, the effect on expected utility maximization is similar to the effect of an increased interest or discount rate. If the discount rate is 5% and the probability of stock collapse is 2% each year, the economic optimum for the fish stock would be like the risk-free optimum but using a 7% discount rate. Similar kinds of risk might arise from market phenomena. Some fisheries are strongly dependent upon maintenance of prices in regional markets. Alaska sockeye salmon producers, and numerous other North American and Asian fisheries, are heavily dependent upon the Japanese market, for example. If there exists a probability that an international trade war, exchange rate shift, or alternative source of supply could suddenly make the market unavailable or unprofitable to the fishery for extended period of time, then this risk of future economic collapse would dilute the perceived future economic benefits of stock conservation.
A different source of risk is stock collapse precipitated by over fishing. In this case the probability of collapse is an increasing function of harvest. Intuition is less reliable here, but I think it is likely that in most such circumstances, the optimal risk averse economic strategy would dictate a lower exploitation rate for any given stock biomass. The logic is not very abstract. If speed increases the likelihood of a costly or disastrous accident, slowing down is both intuitively and economically a good strategy. Environmentalists are likely to support this approach.
One difficulty in using economic models and consequences as a basis for judging the outcomes of alternative management strategies is that there are several, connected, inconsistent measures of economic benefit. Agricultural economists have for decades confronted the problem that policies which stabilize farm prices differ from those which stabilize farm incomes and both of those differ from policies which optimize consumer benefits of farm production or which best protect water quality. In the formal decision model the question of who is being served by the policy is subsumed in the formulation of the preference function. In practice, establishing economic objectives is a participatory process that resists simple description by convenient mathematical functions. This characteristic of decision processes again suggests that technical analysis be focused on providing quantitative descriptions of outcomes under alternative policy regimes. Interaction with the policy makers can help to make the analysis relevant from the perspective of both the outcomes to be evaluated and the policies to be considered.
22 Presentation to the Fishery Management: Global Trends Symposium, June 16. Seattle, WA.
We are bombarded with strident voices objecting to an economic approach to risk management on grounds that it fails to protect future generations, that it takes unconscionable risks. Those working within the decision framework can interpret these concerns in various ways: (1) the preference functions of their critics express more risk averseness, or (2) the expected future losses from risky decisions are being underestimated, (3) the range of feasible actions is not adequately explored, or (4) the risks perceived by the critics are over-stated relative to the technical risk assessment. The first interpretation objects to the preference ranking function, to the weight placed on security or safety. Some people are more inclined than others to avoid risks and to seek security. The second interpretation concerns mainly the quantification of potential losses. For example, when the commercial fishery causes the decline in benthic organisms or marine mammals, this is considered a minor side-effect by industry participants, but is a major negative outcome in the minds of many others. This is essentially the source of conflict that fueled the campaign leading to the United Nations ban on high seas drift net fishing.
The third possibility is that the objectors to technical risk management see feasible solutions where others see none. In western Northwestern America, there is a continuing crisis over extinctions of salmon stocks. Supporters of salmon protection and recovery see relatively painless and practical solutions that focus on habitat protection and modest harvest reduction. Others see these solutions as unproved, unwise, and exorbitantly expensive. As in the safe minimum standard approach to endangered species decisions, the protectors of salmon assume that we can “choose species preservation”. Opponents see this as practically impossible and the attempt to accomplish it as harmful to their interests. The fourth interpretation flows from the technical versus perceived risk discussion. Where some see looming catastrophe, others see manageable risks. Further, as noted earlier, people tend to underestimate familiar risks and to overestimate unfamiliar risks. There seems a particularly intractable division between those who view natural systems as finely-tuned and fragile versus those who see nature as robust and adaptable. I might add that this division extends into the social and economic realm. Some view with great alarm any changes in fishing method, economic power, and related social organization, while others see adaptation and change in technology, social organization, and the laws as normal and desirable.
All four of the issues described above concern controversial aspects of risk management that cannot, in principle, be solved by technical analysis. Because people differ, and because their views change with experience, risk communication and participation in negotiating sessions are necessary to select management measures.
In practice it is difficult to determine objectively whether a rational, utility-maximizing decision process is “cautious”. The choice of actions depends upon attitudes towards risk (risk aversion), the perception of risk (risk assessment), and perceived outcomes of alternative actions (system models). Decision makers incur costs in the form of lower average economic returns in order to avoid other negative outcomes. As indicated above, even a risk neutral person will rationally anticipate the likelihood of negative effects and take actions to avoid negative outcomes so long as the costs do not exceed the benefits. So, risk aversion is not necessary to motivate actions that avoid large negative outcomes. How could one determine whether fishery managers are cautious? Suppose there are a thousand marine fish stocks, that each is susceptible to both natural fluctuations and exploitation, and that one stock crashes each year. Is that too little precaution, or too much? One answer is that is depends on how the risks of collapse are related to the benefits of fishing. Another answer is procedural; if the technical decision information did not adequately display the degree of risk involved, then the decisiona were likely inadvertently risky. The issue of adequate precaution has to be evaluated against a background level of stock variation and collapse that occurs in the absence of fishing. It is also crucially dependent upon whether the collapses are irreversible or simply inconvenient and temporarily disruptive to the human economy.
The discussion of risk assessment and economic concepts in decision making shows that there are ways to incorporate a variety of considerations in formulating precautionary fishery management policies. Among the important characteristics of decision making under uncertainity are:
The process of formalizing technical advice to managers must focus on estimating the degree of risk, establishing realistic alternative management strategies, and gauging the magnitudes of biological and economic losses associated with those risks.
The technical risk assessment procedures need to be combined with risk communication processes in order to reconcile or negotiate significant differences between technical and publicly perceived risks.
Rational economic harvest policies exhibit a degree of precaution, incorporating safeguards against stock collapse, commercial extinction, market price collapse, social instability or whatever outcomes are considered to be “bad”. Whether such a policy satisfies the more extreme demands for safety in stock conservation and ecosystem stability is unclear, since it remains to be shown that representative values and risk attitudes of the affected public would give heavy weight to safety.
I conclude with the following observations.
First, although the formal analysis sketched out in this paper seems complicated, it may admit of simple, intuitive strategies for action. The basic notion of precautionary action -- to proceed cautiously to avoid catastrophic results -- may be given fuller expression through the development of rigorous models. But the final results of that research -- suggested rules of conduct -- need not be complicated. An example is the economic reference point biomass, which can be roughly calculated from rudimentary cost and fish value information. Analogous biological reference points are analyzed in the technical fisheries literature.
Second, it is essential to include economic and social considerations in the formulation of management strategies. In a rational decision process, economic objectives do not make fishery management more risky. However, the interests of participants in a free access fishery tend to be myopic and short-run concerning fish conservation. If these economic interests dominate the decision process, conservation practices may suffer as the desperation of individuals under economic stress makes precaution a distant concern. Hence, institutions matter greatly, in particular those property rights institutions that create longer term interests and security of tenure to fishing people. This does not mean that individual property rights, like individual fishing quotas, need to be implemented in every fishery. Institutions that strengthen the hold of fishing people over the fish stocks on which they depend, however, should increase the receptivity of fishing interests to precautionary measures.
Third, although I have suggested economic and social concerns can be considered in risk assessment and in the decision process, this does not mean that extensive social science research is necessarily the path to better management. The stochastic modeling approach already developed by fishery modelers could be expanded to test whether precautionary approaches are significantly different with explicit social and economic objectives. That research would suggest specific social-economic research projects that could improve decision processes. How such information can contribute to better management of fishery risks depends upon particular contexts and needs to be determined case by case.
If precautionary biological reference points are roughly equivalent to economic reference points, there may be little to gain from the added complexity to management analyses inherent in multiple objectives. The greatest contribution of the economics is often a deeper understanding of the main consequences of risk and of the importance of the role of risk in human organization and action. But the selection and implementation of conservation objectives needs to incorporate the risk communication and negotiation processes mentioned earlier. These processes provide means to incorporate public perceptions and ancillary concerns, such as costs of enforcement, effects of regulations on social institutions, and economic impacts. The collaborative process tends to expose the socially acceptable forms of fishery intervention, to expand the menu of organizational responses to risk, and to reconcile differences in values and risk perceptions between fishing groups and between technical analysts and the general public.
Anderson, L. G. 1986. The Economics of Fisheries Management. Johns Hopkins: Baltimore.
Bishop, Richard C. 1978. “Endangered species and uncertainty: the economics of a safe minimum standard,” American Journal of Agricultural Economics, 60(1):10–18.
Bishop, Richard C. 1993. Economic efficiency, sustainability, and biodiversity. AMBIO, 22(2–3):69–73.
Clark, Colin W. 1973. The Economic of overexploitation. Science 181:630–634.
Clark, Colin W. 1975. Mathematical Bioeconomics. New York: John Wiley.
Clark, Colin W. and Gordon R. Munro. 1975. The economics of fishing and modern capital theory: a simplified approach. Journal of Environmental Economics and Management 2: 92–106.
Finkel, Adam M. and Dominic Golding (Ed.). 1994. Worst Things First? The Debate over Risk-Based National Environmental Priorities. Resources for the Future, Washington, DC.
Freeman III, A. Myrick. 1993. The Measurement of Environmental and Resource Values. Resources for the Future. Washington, D.C.
Hilborn, Ray, Ellen K. Pikitch, and Robert C. Francis. 1993. Current trends in including risk and uncertainity in stock assessment and harvest decisions. Can. J. Fish. Aquat. Sci. Vol. 50: 874–880.
Hirshleifer, Jack and John G. Riley. 1979. The Analytics of Uncertainty and Information -- An Expository Survey. Journal of Economic Literature Vol. XVII (Dec.), pp. 1375–1421.
Hirshleifer, Jack, 1992. The Analytics of Uncertainty and Information. Cambridge Surveys of Economic Literature. Cambridge University Press, Cambridge, UK.
Kahneman, D., Paul Slovic, and A. Tversky, eds. 1982. Judgment Under Uncertainty: Hueristics and Biases. New York: Cambridge University Press.
Knight, Frank H. 1921. Risk, Uncertainty and Profit, New York: Houghton Mifflin.
Krutilla, John V. and Anthony C. Fisher. 1975. The Economics of Natural Environments: Studies in the Valuation of Commodity and Amenity Resources. Baltimore: Johns Hopkins University Press.
Lee, Kai N. 1993. Compass and Gyroscope, Integrating Science and Politics for the Environment. Washington D. C.: Island Press.
Leiss, William and Christina Chociolko. 1994. Risk and Responsibility. Montreal: McGill-Queen's University Press.
MacCrimmon, K. R. and D. A. Wehrung. 1986. Taking Risks: The Management of Uncertainty. New York: The Free Press.
Merkhofer, Miley W. 1987. Decision Science and Social Risk Assessment. Dordrecht, Holland: D. Reidel Publishing Co.
Mitchell, Robert C. and Richard T. Carson. 1989. Using Survey to Value Public Goods: The Contingent Valuation Method. Washington, D.C.: Resources for the Future.
Montgomery, Claire and Gardner M. Brown. 1992. Economics of species preservation: the spotted owl case. Contemporary Policy Issues X(April): 1–12.
Neumann, John von and Morgenstern, Oskar. 1944. Theory of Games and Economic Behavior, Princeton, NJ: Princeton University Press.
Savage, Leonard J., 1954. The Foundations of Statistics. New York: Wiley.
Smith, Stephen J., Joseph J. Hunt, and Denis Rivard.1993. Risk Evaluation and biological Reference Points for Fisheries Management. Can. Spec. Publ. Fish. Aquat. Sci. 120.
Smith, V. Kerry and John V. Krutilla. 1979. Endangered species, irreversibilities, and uncertainty: A Comment. American Journal of Agricultural Economics 61(2): 371–375.
Suter, Glenn W., III. 1993. Ecological Risk Assessment. Chelsea, Michigan, USA: Lewis Publishers.
Thompson, Grant. 1993. A proposal for a Threshold Stock Size and Maximum Fishing Mortality Rate. In Smith, Hunt, and Rivard. Risk Evaluation and biological Reference Points for Fisheries Management. Can. Spec. Publ. Fish. Aquat. Sci. 120. pp 303–320.
Walters, Carl. 1986. Adaptive Management of Renewable Resources, New York, McMillan Publishing Co.