Department of Agricultural and Resource Economics

University of California

Paper Presented at:

Heartland Environmental and Resource Economics Symposium

Ames, Iowa

September 17-18, 2000

I have frequently been asked: "What is environmental economics?" (The truth is the same was asked about agricultural economics.) How does it relate to economics or the environment? Does it have its own theory? Is it basic or applied? I would like to provide here my personal perspective and illustrate it with examples that mainly relate to the economics of conservation and the management of environmental risks. This perspective is colored and biased by my experience as an agricultural economist, and I think that the basic methodological foundations of agricultural and environmental economics are very similar. I am engaging in this self-examination of our discipline because I believe that it is needed from time to time, and it is important for our focus, productivity, and self-esteem. I conclude that environmental economics has its own theories and basic lines of research that are unique and challenging. When practiced appropriately, theoretical and analytical frameworks of environmental economics are unique, interdisciplinary, and integrate the core tools of economics with findings and models of the natural sciences to obtain original insight and policy perspectives.

Economics has a stylistic and generic depiction of the world, expressed by its assumptions. Its theory is developed by using mathematical tools to provide insights on allocational choices by individuals; behavior, and performance institutions, in particular, markets; and the realization of several important variables including prices, outputs, inputs, incomes, etc. Economics has an ingenious capacity to distinguish between what is happening (positive analysis) and what should happen (normative analysis) and thus identify the need for policy interventions. Econometrics and other techniques of applied economics modified statistical methodologies to estimate and compute the many concepts and relationships invented by economic theory. All the qualitative and quantitative results of economic analysis have had an immense impact on our society and provided the foundation for policymaking in many areas. Yet, economists still have the "Rodney Dangerfield complex." They perceive that in many cases they get no respect. In many cases, politicians and the public reject ideas of economists because of political and self-interest reasons. In other cases, policymakers may reject economic advice because of lack of understanding or education. Sometimes our analysis does not seem realistic because it may obstruct many important aspects of reality, and because of the tendency of many economists to be isolated from other disciplines, self-contained, and self-sufficient.

Obviously, I am oversimplifying here, but it seems that in developing theoretical framework, the core of economics has a close reliance on mathematics. In recent years we have seen a welcomed interaction with political science, psychology, and sociological that have led to some new and creative models (see, for example, Akerlof and Robin). Empirical economic analysis relies, as I mentioned before, on basic principles of statistics and "the available data." Economists attempt to use primary or aggregate data and raw materials and estimate some of the key economic constructs without investigating or incorporating knowledge of other disciplines on properties of some of the data or the processes that generated it.

The tendency of economists to be excessively self-reliant is manifested by the use of duality-based relationships for estimation of technological parameters (Chambers). This ingenious approach will enabling obtaining key parameters of production functions, the output supply, and input demand using mostly monetary data (revenues, costs, prices), and some physical data on input and output to estimate the parameters of production functions, output supplies, and input demands. These estimated relationships are very valuable for productivity comparisons, to assess the response to some changes in monetary incentives, etc. However, the parameters that are derived using this approach are not pure technological relationships. Their estimation procedure incorporates the assumptions of profit maximization (or expected utility maximization). Thus, the estimators of what is supposed to be a technological parameter (e.g., elasticity of substitution between labor and fertilizer) may change with changes in our behavioral assumptions. Since duality-based estimators are hybrids of technological and behavioral assumptions, they are limited in their capacity to indicate to what extent the firms take advantage of technological opportunities. They may be of limited use when one wants to consider implications of new regulations or incentives that aim to modify behavior to address an environmental constraint. In our work on the impact of environmental regulation or water policy, we realize that we would not be able to obtain realistic estimators unless we interact and incorporate the knowledge of technological experts, engineers, agronomists, farm advisors, and the farmers or businessmen themselves.

As economists, we preach the merits of relative advantage and gains from interaction and trade and also conduct policy-relevant research on the impact of environmental policies and regulations. Physicists, engineers, and biologists have obtained key relationships that underline production and pollution technologies. capabilities. Incorporating them into our economic optimization will add extra structure and improve the relevance and realism of our results.

Lancaster’s original criticism of demand theory applies to a large extent to production economics. Lancaster argues that traditional demand theory that culminated in the Slutski equation was elegant, but rather limited. For the sake of generality, it shuns many assumptions, but the result was a small number of concrete predictions. He suggested that we should expand the structures of our demand and consumer behavior models, and his insights led to the development of hedonic price model, family production function models, etc. Recently in a survey article on the state of production economics, Just and Pope reached a similar conclusion. Our models of production on the one hand are restricted by behavioral assumptions, and on the other hand lack technological specifications that significantly limit their usefulness. To be an effective player in environmental policy debate, economists have to be well informed and analyses have to be as specific and detailed as possible. Incorporation of basic scientific principles and relationships to economic models is a first step in upgrading our analysis and contributions to environmental policy.

Resource economics has already reached interdisciplinary traditions. Much of the analysis of fisheries has been developed by a mathematical biologist, Colin Clark, and the classical empirical study of fisheries by Crutchfield and Zellner integrated estimation done by population biologists within an economic decision-making framework. The Faustmann model of forest management, which provides the starting point to much of forestry economics starts with modeling the dynamics of tree growth. I will demonstrate below how simple physical relationships can improve the analysis of externalities and environmental health.

Material balance equations are simple, physical accounting equations that state that the sum of inputs entering process is equal to the sum of output coming out of it. These equations apply to water, energy, chemicals, and other objects of concern in environmental policy. It is important to recognize that in many processes not all the inputs that are applied are actually consumed by the process to provide the desirable output. There is a gap between the effective input used in the production processes and applied input, and this residue in many cases is a source of environmental problems. Kneese and Ayres recognize the importance of material balance equations and suggested that environmental economic analysis will integrate monetary and physical accounting to have a full accounting of a production processes which lead to full pricing and resource allocation that will recognize the environmental cost and benefits of certain activities. Engineers and agronomists have developed notions of technical efficiency (e.g., energy efficiency of an appliance, water use efficiency of an irrigation system) to evaluate alternative technologies and management techniques.

Khanna and Zilberman provide a list of examples where externality problems are originated in unutilized inputs. This is the case with many water quality problems associated with application of chemicals such as fertilizers and pesticides. Unburned fuel in cooking processes may evaporate and be a major cause of air pollution in developing countries. Sprayed chemicals may stray and damage unintended targets, etc. It is essential to incorporate the material balance equations when they apply in economic models of pollution and externality problems. When pollution is a residue, activities that increase output per unit of input are consistent with reducing pollution intensity. Thus, the assumed tradeoff between pollution and output does not always hold and in many important cases, there is actually complementarity between output and pollution. As Caswell et al. and more generally Khanna and Zilberman suggest, improving input use efficiency and reducing pollution intensity may entail extra cost in terms of investment in conservation technology or management practices. However, ignoring the complementarity between environmental quality and output, when it exists, may lead to wrong policy prescription. If for example, a policymaker who cannot observe pollution uses a tax as a mechanism to reduce pollution (as suggested by many authors), the net result may be to discourage "cleaner" producers who adopt the technologies that increase input use efficiency and reduce pollution intensity. That suggests that in cases where residues generate pollution and pollution cannot be monitored, taxation input is preferable to taxation of output.

This analysis sheds more light on the role of technologies and incentives to manage technological change in pollution control. Khanna and Zilberman interpret the notion of precision technologies "generically," namely, these are technologies that apply input whenever they need it as thus reduce residues and pollution. Caswell et al. use modern irrigation technologies as examples of such technologies which adoption can significantly reduce drainage problems in California while increasing yield. The new communication and electronic evolution developed new capacities of monitoring situations and refining applications to reduce, say, drift of pesticides or chemicals and to target inputs to where they are needed. Improvements in computer technology are likely to lead to significant improvements in the efficiency and reduction in pollution of transportation systems. But introduction and adoption of more precise but costly technologies will not always occur without incentives. In many cases, the gain in output associated with increased precision may not pay for the extra cost and without taxing of the pollution or input related to the pollution or even subsidization of adoption to account for the gain in terms of environmental quality, adoption would not occur. Thus, when we look at the role of environmental incentives, we don’t need only to consider the short-term effect of reduction in use of polluting inputs within a given technology, we should also consider the effective of new innovations and adoption of cleaner technologies.

This last distinction is consistent with a broader perspective of technology that is presented by the putty-clay approach of Johansen and Salter that I find very useful. Individual production units (machines) tend to have rigid production coefficients; namely, the variable inputsto output or pollution to output ratios of a given piece of equipment are rather fixed. Of course, different pieces of equipment at different locations have different coefficients. Producers may reduce its pollution levels by either stopping utilization of existing equipment or by investment in new equipment or significant modifications of existing ones. That suggests that in the short run environmental policies that will penalize polluting units may reduce output because the major response will be to shut out polluting units. However, in the long run they may increase output, as people will adopt new technologies that increase input use efficiency and thus simultaneously increase output and reduce pollution.

Here I will add my interpretation of the Kuznets curve. The extra cost and sophistication associated with the development and adoption of precision technologies suggest a dynamic evolution that resembles the Kuznets curve. In early stages of the modern era, countries in the West discovered new technologies that utilized forms of viable inputs such as fossil fuel or chemicals that increase productivity but generated pollution. The use of this input was first with technologies of low precision and relatively low input use efficiency. Over time, due to increased concerns for environmental quality and the resulting environmental regulation and increased scientific knowledge, more precise technologies are being introduced. These technologies continue to enable economic growth but with lower pollution intensities.

One of the conclusions of studies of input use intensities is that they are dependent on environmental conditions and technology choice. Traditional gravitational irrigation technologies have input use efficiency of .9 on heavy soil and of .4 in sandy soil, so that the gain from adoption of a modern technology in heavy soil may be minimal while in the sandy soil it may be substantial because modern technology may have input use efficiency of .9. Similarly, the drift of chemicals from aerial spraying is relatively much larger in locations which are subject to strong winds than in locations with calmer weather patterns. That suggests that, when we conduct environmental policy analysis, we have to recognize the importance of heterogeneity across locations. With heterogeneity, we are not likely to find one technological solution that fits all locations, and it is very important to develop technological solutions that are appropriate for environmental and socioeconomic conditions.

Quite frequently, economic analysis for the sake of simplicity assumes homogeneous economic agents (consumers, producers, etc.). Assumption of homogeneity does not hold in many cases for various areas of economics. However, it is particularly unrealistic when it comes to environmental problems. Economic agents may be heterogeneous both in terms of the factors that affect profitability of their main activities as well as key parameters that determine environmental side effects of their production or consumption choices. Therefore, modeling tools that link technological choices to key economic and environmental parameters and information about the distribution of these parameters over space are especially valuable. That information will enable obtaining quantitative assessment of changes in technology, production, and pollution levels in different locations in response to changes in policy, as well as aggregate measures of impacts by aggregation over space.

We have shown that recognizing the differences between applied and effective inputs and the role that technology and the physical conditions play in determining input use efficiency at various locations may be essential for analyzing pollution problems that were caused by unutilized residues of applied inputs. In many cases, the link between pollution and health is not that obvious, and some of the scientific knowledge that may be useful to analyze processes of environmental health risks is discussed in the next section.

The ultimate motive for many environmental policies and regulations is protecting human and environmental health. Therefore, effective modeling and analysis requires good health measures and conceptual and quantitative understanding of the relationships between various activities of economic agents and indicators of health. The public health literature has used various measures of risk as indicators of the severity of human and environmental health problems. The measures and definitions of risks used in public health are intuitive but rather different than measures of risks used by economists. The risk associated with a certain activity, say, pesticide use is the probability of bad outcomes (death or disease) for a specific individual (or a random member of a population) during a certain period of time because of the activity. There may be several measures of risk associated with the same activity. It can be measured in terms of mortality or occurrence of a disease (when there is concern for a disease, an alternative measure is sick days). Because populations are heterogeneous and some members are more vulnerable to certain activities than others, it is useful to distinguish between types of risks. We may sometimes be concerned with acute risk (e.g., probability of death or severe injury or short periods of time after an exposure), while in other situations the main concern is with chronic risk (e.g., probability of cancer that may occur several years after exposure or after a long period of exposure).

There are two different, yet relative, procedures to estimate the relationship between activities and risk. One approach that has been used frequently by epidemiologists is conducting statistical analysis and relating incidences of death at different times to environmental conditions, human activities, individual characteristics, etc. A more analytical approach, which is especially applied to low risk problems such as risk associated with pesticide use, is introduced by a methodology called risk assessment (Bogen). This approach models risk as outcomes of several processes. The risk associated with a certain pollutant depends on the magnitude of the pollution, processes of transport and fate (since pollutants may move and maybe transformed to other materials), exposure, and dose response. It is quite common to use a multiplicative specification for risk-generating processes so that risk is a product of pollutants times transport and fate coefficients times exposure coefficients times dose response coefficients. It is very useful for economists and policy analysts to think in terms of the risk assessment models because each of the processes that generate the risk is affected by policy interventions. Taxes, direct control, etc. can affect the level of pollution. The process of transport and fate can be affected by erecting barriers. Exposure can be affected by installing filters, requiring the use of protective clothing, etc. Dose response, which indicates vulnerability to exposure, can be modified by medical treatment. Furthermore, the coefficients of the risk generation process are subject to a high degree of variability. This variability may be caused by three phenomena:

1. Randomness. Weather conditions, for example, significantly affect the process of transport and fate.

2. Lack of knowledge. This reflects the fact that we use only estimates of the key parameters of the risk generation process, and these parameters have a high degree of uncertainty.

3. Heterogeneity among the populations.

The estimates of risk have a high degree of variance, but this variability can also be affected by the choice of policies. Thus, every policy intervention has an effect both on expected risk as well as variability of risk. In addition, extra research is another policy option is that will increase the knowledge of policymakers about the risk-generating process and the reliability of estimates of risk. In choosing policy interventions, policymakers consider three elements: the impact of the intervention on both the expected value and variability of risk and the cost of the policy intervention. In some situations, the most effective intervention may be a source control, namely, an incentive or regulation to reduce pollution. However, in other cases, when the cost of pollution reduction is high or the ability to monitor it is very low, intervention may have a different form. Sometimes policymakers may use several types of intervention. They may introduce regulation or incentives to reduce application of toxic materials, require individuals who are likely to be exposed to wear protective clothing, and develop medical emergency procedures in case of exposure.

Lichtenberg and Zilberman introduced a framework for environmental regulation that explicitly incorporated the risk assessment models. They considered situations where policymakers are interested in containing risk with a certain degree of reliability at minimum cost or have to allocate a given budget to reduce risk, again, subject to the degree of reliability. Their numerical analysis shows that reliability matters significantly. The cost of controlling the risk of cancer from drinking water that may have been contaminated with DBCP with a 99 percent reliability is twice as high as the cost of containing the risk at the same level with a 95 percent reliability. Optimal policy intervention is determined where the value of the marginal effect of policy on risk is equal to the marginal cost of the policy intervention, and the policy levels vary with the degree of reliability of the risk contained. Obviously, stricter policies will be imposed to reduce the variability of risk.

It is important to recognize that risk assessment studies may produce estimates of risks derived under varying degrees of reliability. Using the same data, one study may estimate the risk to be, say, $1,000 and the other $1 million. The difference may be that one study will use the mean of the risk distribution as the estimator of the risk, while another study may use the 99 percentile of the community risk distribution. When the mean is used as an estimator of the risk, then it conveys a much lower risk level then the 99 percentile of the risk distribution. Thus, scientists who may work for producers (and polluters) may use a mean risk as an estimator of risk, while scientists who work for an environmental group may use the 99 percentile of the cumulative risk distribution. The lack of established criteria for determining the exact risk estimator if one number is used as an estimator might cause significant confusion in policy analysis. Lichtenberg and Zilberman argue that the first step in developing the capacity to compare environmental policies that affect risk is to use estimates of risk that have the same degree of statistical reliability. It does not matter if policymakers use the mean of the risk or the 99 percentile of the risk distribution as a risk indicator as long as the measures of risk have the same degree of statistically reliability.

When the policy objective is to minimize the cost of reaching a certain risk level with a certain degree of reliability, then the optimal policy implies a shadow price for the risk. This shadow price can be interpreted as the cost of saving a statistical life. Environmental and health regulations are consistent if the implied shadow prices are consistent across policies. Cropper et al. found a high degree of variability of implied values of life associated with pesticide policies, as well as other environmental policies. However, it is not clear to what extent this inconsistency is due to inconsistent policy process or the fact that in many studies the risk estimators were derived under different degrees of statistical reliability. Probably both inconsistent policymaking, as well as inconsistent measurement and estimation criteria, resulted in significant variability in the estimated value of life associated with different policies. Thus, it is important to have consistent risk estimation procedures and consistent statistical reliability as we strive to develop consistency in policymaking to affect environmental regulations. As economists, we can raise these issues of measurement, inconsistency, and demand when we obtain policy-relevant information from other disciplines, and this element of consistency will ensure a smoother policymaking process. However, we shouldn’t be content with just getting the final estimate of risk with certain activities. We should, instead, obtain an understanding of the risk-generation process because, as we said before, understanding such a process will provide a much richer set of policy option selection when examining the best combination of risks as well as economic well being.

We have shown that incorporating biological and physical relationships and knowledge into economic decision-making framework provides a new insight on producer behavior and, in particular, on technology choice and determination of pollution and output levels under alternative policies. Our discussion of risk assessment models suggests that, when scientific knowledge is added to economic models, it expands the range of policy levers to be considered, improves our interpretation of environmental and health indicators, and leads to better understanding of the uncertainty associated with environmental modeling and decision making.

The integrated policy modeling approach we presented here should apply to most environmental policy problems. Pest control issues have been a major source of concern since the publication of Silent Spring, and the framework for their regulation should encompass agronomical and ecological information relating to pest dynamics and productivity. Public health models address the environmental and health effects of pesticide residues and economic models of producers’ behavior and pesticide markets (NRC; see applications in Harper and Zilberman and Sunding and Zivin). Studies on the impact of climate change, say, on the agricultural sector, has to incorporate natural science knowledge and models on the direct effect of changes in temperature on productivity. It also has to consider the fertilization effect (the impact of changes in carbon concentration in the atmosphere on productivity), the impact of changes in weather patterns on pest movements, the impact of climate change on precipitation and water resources, and economic decision making that recognizes spatial heterogeneity. It also has to incorporate the specific feature that characterizes the interaction of the public and private sector in agriculture, both in terms of support programs as well as in terms of the provision of research and development. Developing a comprehensive model that will include everything may be difficult or even impossible but developing several complementary modeling frameworks, with each emphasizing specific aspects and then linking their results together, may be feasible and beneficial and result in a better understanding of policy analysis.

The message of this paper is that environmental economics is a distinct discipline, where the environment and economics are integrated rather than adding environmental issues to economic frameworks. Environmental economists should know about the key features of the system that they are dealing with and incorporate some of the main results and models. They should develop working relationships with noneconomists who can provide key insights and information. They should be aware of real-world policy issues and dilemmas and how they apply to our key assumptions.

While working with noneconomists over the last 20 years, I have learned that some of them think like economists and actually have come up with wonderful economic insights. Some of the models of ecologists and public health specialists and biologists can be easily transformed and included into economic decision-making frameworks, and the models may provide us with wonderful data for applications and case studies. On the other hand, in many cases some of the basic knowledge or data needed for economic analysis may not be available to scientists, and we as economists can trigger research in biology, physics, or agronomy that will result in output that will be policy relevant.

I perceive that there is asymmetry in the way scientists perceive knowledge in their discipline versus other disciplines. As economists, we are aware of the limitation of our knowledge and our data. We know that we may not have the answers to many important questions that outsiders may expect economists to solve. While at the same time, we may expect that scientists have the answers to what we consider to be technical questions. We expect engineers to know how inputs relate to output and require public health specialists to have quantitative knowledge on the relationships between certain activities and environmental conditions. Obviously, that is not the case. There are gaps in biological and physical knowledge. The research agenda of biology and physics is not designed to answer questions that are of interest to economists and policy analysts. While there is a lot that can be taken and adopted from the other disciplines, our interaction with them may trigger new lines of research and provide better information that will help us in the future. The role of environmental economists in the policy process is to be not only an integrative discipline that provides balance and the rational policy answers but also to raise new research issues that other disciplines can solve and to improve the decision making in the future. However, I don’t see that as a one-way interaction. As we interact with other disciplines in developing an environmental economics agenda, we may gain knowledge and also discover new questions that we should ask.

I would like to summarize the lesson I have learned about environmental economics and its role with a metaphor. As we view ourselves within the academic sky, I don’t see environmental economics as a satellite of economics but, rather, I view it as an independent entity in the galaxy of science. It is very close to economics, but it is vitally influenced by other disciplines and has methodologies, knowledge, and values of its own.

Khanna and Zilberman provide a list of examples where _____________ problems are originated in unutilized inputs.

Kneese and Ayres recognize the importance of material equations and suggested that environmental economic analysis will integrate monetary and physical accounting