Explaining Environmental Policy Failure

A proposed new approach to environmental policy fails to convince.

Why Environmental Policies Fail by Jan Laitos is a curious book (1). Its subject is certainly important, and it’s hard to disagree strongly with the bleak picture it presents of risks to the environmental conditions needed for humans to live safely  (pp 59-76).  It contains interesting discussions of particular situations and policy failures.  What makes it curious is the abstract framework within which its proposals are presented, based on what it describes as “fundamental laws of nature” (p 171).

The author is a professor of environmental and natural resources law, and it’s interesting  to see how he pigeonholes what economists regard as two key policies to address environmental issues: Pigovian taxes and marketable permits.  Grouped within the heading of economic policies, these form just one of ten items on his list of current environmental policy strategies (pp 139-140 & 145).  Among the others are some which are commonly discussed by economists as alternatives to market-based instruments, eg regulations (command and control), and adjustments to property rights.  Some others are new to me: legal rights for nature (pp 153-9) and human-nature linkages (pp 160-3).

Laitos’ comments on all these policy strategies are generally critical.  In the case of economic policies, his main criticism is that they are based on assumptions that human behaviour is deliberate, rational and utility-optimising. He refers to evidence that people often behave quite differently (pp 32-3, 145-8). But a crucial question, which he does not consider, is how much difference this makes.  Consider a case where emissions of pollutant are initially at a rate of 100 units per day.  Suppose that, applied to relevant data, conventional economic analysis predicts that a proposed tax would reduce the rate to 50 per day.  Given a more realistic view of behaviour, we would not expect the outcome of the tax to be exactly 50 per day.  But will it be, for example, 55 per day or 90 per day?  The former would suggest that the theory works fairly well in this situation (perhaps because departures from rationality largely offset each other in aggregate); the latter that it does not.

Environmental Policy Failure – an Orthodox Account

Laitos’ overall assessment, as the book’s title suggests, is that environmental policies to date have been unsuccessful.  His argument is essentially that severe environmental degradation has occurred despite the presence of many environmental policies (p 61).  I will not dwell on the objection that some policies, for example the Montreal Protocol on Substances that Deplete the Ozone Layer, have been fairly successful (2).  Of more importance here is that he shows little interest in distinguishing types of failure.  The following is I suggest a useful classification:

Type 1 Failure: Environmental policies which fail to achieve their objectives.

Type 2 Failure: Environmental policies which produce undesirable environmental side-effects.

Type 3 Failure: Environmental policies for which the objectives might be considered insufficiently ambitious.

All three types occur (some policies may exhibit failure of more than one type).  As examples I offer the following:

Example of Type 1 Failure:  The US Lead and Copper Rule, intended to minimise lead and copper levels in drinking water (3), which did not prevent lead contamination in Flint, Michigan (4).

Example of Type 2 Failure:  Germany closed eight of its seventeen nuclear reactors in 2011 following the Fukushima disaster in Japan (5).  This significantly reduced its exposure to the environmental and safety risks of nuclear power.  But to ensure adequate electricity supply, Germany has since increased production of brown coal which is an even worse source of air pollution and greenhouse gas emissions than ordinary coal (6).

Example of Type 3 Failure: Australia’s greenhouse gas emissions target for the Kyoto Protocol’s first commitment period was an increase of no more than 8% on its baseline level (7).

The reasons for these different types of failure are likely to be quite different.  Type 1 failures are often due to poor technical advice to policy-makers and/or ineffective administration and enforcement (including regulatory capture by rent-seeking groups).  Type 3 failures tend to result from political compromises between differing interest groups, or between short and long term interests.  Type 2 failures may occur for either of these reasons, or because a particular environmental problem happens to be especially salient when a policy is adopted.  We have, therefore, the outline of an orthodox answer to the question implicit in the book’s title.

Environmental Policy Failure – Laitos’ Account

Laitos, however, largely neglects these sorts of explanation. Instead, he presents a framework of thought, the key elements of which can be summarised as follows (8):

Nature as a Complex Adaptive System, not necessarily in equilibrium, characterised by co-evolution of living organisms and their abiotic surroundings, and often achieving resilience between stability and instability in the state sometimes termed ‘the edge of chaos’ (pp 95-100).  This leads to the idea that environmental policy should not aim to conserve or restore a state of nature, because there never was such a stable baseline state (p 100).

The Requirement of Symmetry as fundamental to the laws of nature, in physics and in biology (p 174).  The following three consequences of symmetry are of particular relevance to environmental policy.

The Law of Conservation according to which certain quantities cannot be created or destroyed, an example being the First Law of Thermodynamics  (p 177).  This leads to the conclusion that, because the natural environment is a complex adaptive system, environmental policy should be simple, and complex policies are likely to fail (p 179).

The Equivalence Principle according to which seemingly different concepts are actually the same, an example being the equivalence of gravity and acceleration in Einstein’s general theory of relativity (p 180).  This leads to a rejection of the ideas that humans are morally superior to their environment, and have the ability to manage and master it.  Instead, environmental policy should treat humans and their environment as equivalent parts of a social-ecological system (p 181).

The Unification Principle according to which apparently opposite concepts are actually the same (pp 181-2.  Again, there are examples from physics.  The conclusion drawn is that environmental policy should address not simply the environment but the whole social-ecological system (p 183).

According to Laitos, therefore, environmental policies do not work well for four main reasons: they are based on an inaccurate model of nature; they often presuppose human superiority; they are inconsistent with the law of symmetry; and sometimes they are based on a false model of how humans behave (p 184).

The most persuasive part of this is the view of nature as a complex adaptive system.  We know for example that there are complex interdepencies between the biotic and abiotic elements of the environment, via processes such as photosynthesis and carbon sequestration.  We also know that such natural processes have not always been in equilibrium.  The oxygen content of the atmosphere, 21% at present, has during the last billion years been as low as 3% and as high as 35% (9).  Even within the much more recent past, there have been dramatic changes such as the retreat of glaciers at the end of the last ice age about 12,000 years ago (10), and a period about 10,000 years ago when much of what is now the Sahara desert received sufficient rainfall to support a savanna-like environment (11).  Certainly, therefore, environmental policies should be developed with awareness of the complexity of natural processes and the possibility that the past may not be a good guide to the future.

On the other hand I am not persuaded by the ‘no natural baseline’ argument. Consider how it might apply to the issue of waste plastic polluting the oceans with adverse consequences for wildlife and indirectly (via food chains) for humans.  The argument would presumably be that the former plastic-free ocean was just one state which had persisted over a certain period and which, given the complexity of nature, might have come to an end without human intervention.  Hence to regard plastic-free oceans as a baseline to be restored if possible would be misguided.  Now that argument seems to me mistaken: it ignores both the implausibility of plastic being produced by natural processes, and the fact that, for thousands of years, fish caught from plastic-free oceans have been an important component of human diets.  What’s more, it illustrates a general shortcoming of Laitos’ approach.  His criticism of actual environmental policies contains many examples of particular policies in particular situations.  But in his presentation of his own policy ideas, I can find little indication that he has tested his abstract principles by considering what they would imply for specific situations.

So far as the requirement of symmetry is concerned, I will not comment on its application to theoretical physics, which is not my specialism.  But I have to say that I find the drawing out of consequences for environmental policy not just unconvincing but in places positively bizarre.  Nevertheless, let us consider the three conclusions on their merits, ignoring their provenance.

Simplicity can indeed be a virtue of environmental policy.  The five pence charge for single-use plastic bags introduced in England in 2015 is an example of a fairly simple policy (12).  It’s easy to think of ways in which the policy might with apparent justification have been made more complex, such as different charges for different bag sizes or exemptions for those on low incomes.  Nevertheless, the policy seems to have been widely accepted.  One likely reason is that the charge is too small to affect anyone’s vital interests.  Another is that it seems obvious that the administrative costs of the sort of complications listed above would greatly outweigh the benefits.

On the other hand, consider the case of international negotiations on reducing greenhouse gas emissions so as to mitigate climate change.  Here vital interests are at stake, along with different understandings of what the scientific evidence shows and different views of fairness in apportioning reductions.  A requirement that any agreement be simple, such as a uniform percentage reduction in emissions, would be fairly certain to ensure failure via a collapse in negotiations. Acceptance that an agreement will need to be complex keeps open the possibility of a partial success – an agreement that if implemented would lead to a reduction in emissions, albeit not as large a reduction as many consider necessary.

Laitos’ rejection of human superiority might suggest that he is an advocate of animal rights.  In fact, the concept of animal rights receives only one brief mention (p 156).  Nor does he ascribe value to nature generally, except in so far as the degradation of nature affects humans (p 38).  The superiority that Laitos rejects is the idea that humans have the ability to manage the environment (pp 29 & 181).  This highlights another weakness of his approach.  Perhaps as a consequence of his high level of abstraction, he tends to make statements that are too absolute and do not allow for matters of degree.  If he had just said that, when trying to manage their environment, humans often fail due to lack of ability, or make mistakes due to lack of knowledge or cognitive errors, then it would have been easy to agree.  But to deny outright  that humans have the ability to manage the environment is too strong.  Human actions do affect the environment, and humans do have the ability to choose among different actions that have different consequences.

If however one really did consider that humans cannot manage their environment, then the natural conclusion would be that any environmental policy is doomed to failure.  It is hard to see how such a belief could be a basis for choosing one policy rather than another.  Yet as we shall see, Laitos does advocate a particular type of environmental policy.

Laitos’ third conclusion – that environmental policy should address the whole social-ecological system – could be taken as just a way of saying that policy formulation should have regard to political feasibility (the social part of the system) as well as to its effect on the environment (the ecological part).  If so it would be entirely acceptable.  But what he actually means by this seems implicit in his rights-based approach to environmental policy, to which I now turn.

Laitos’ Proposed Environmental Policy: A Rights-Based Approach

Laitos proposes that environmental policies consistent with his framework would confer a certain type of right and impose a corresponding duty (p 36).  The right would be:

Positive, that is, a right entitling the holder to the help of others (as opposed to a negative right which merely requires others not to take certain actions) (pp 38-9 & 185-6).

Held by the social-ecological system, that is, by both humans and the environment. Thus it would be quite different from a specifically human right such as a right to clean water (p 185).

To environmental conditions in which humans can live safely (p 185). Thus although the right is held by humans and the environment, its purpose is solely to meet a human requirement, not to save or protect the natural environment (p 187).

Inclusive of the power to alter the right (pp 191).  Thus any legal right granted to the social-ecological system should be capable of being amended (p 192).

Laitos’ account of this right leaves unaddressed the key question of who would grant it.  The answer, at his abstract level, must be humans.  But that leads back to issues of political authority and international negotiation.  For he consistently identifies the right-holder as the social-ecological system. He is not talking about local ecosystems and their human inhabitants, on which one might envisage rights being conferred by local or national governments, but about one system consisting of the global environment and the whole of humanity.

That the right should be capable of being amended seems to amount to saying that there should be flexibility to adjust environmental policies in the light of circumstances (p 192).  As such it seems entirely acceptable.  But again it raises issues of authority which Laitos does not address.  How far should administrators have delegated authority to adjust policies (with risks of uncertainty for those affected and even outright abuse), and how far should changes be a matter reserved for law-makers?

So far as enforcement of the right is concerned, this is best considered after introducing the corresponding duty.  This duty is:

Imposed on humans only, not on the whole social-ecological system (p 185).

Affirmative, that is, a duty to provide something, not merely to refrain from harm (p 197).

To create public environmental goods and positive externalities (p 197).  In this case Laitos does give an example: a person planting a tree which will sequester carbon and so contribute to mitigation of climate change.

Laitos supports the affirmative nature of the duty with the argument that, as a matter of psychology, people respond better to being told what to do than what not do.  This seems problematic.  For one thing, the difference is often just a matter of polite wording, which is important in everyday intercourse (compare “Could you close the door” and “Could you not leave the door open”) but should not be a key consideration in framing laws.  For another, being told what to do can be far more restrictive of individual liberty than being told what not to do (compare requiring a car-owner to travel by public transport, or not to exceed speed limits).  Most fundamentally, the natural wording of some essential environmental policies is negative.  Consider safety policy for nuclear power stations. One might try to formulate a safety policy in terms of ‘preserving a radiation-free environment’, but the more obvious formulation is ‘avoiding radiation leakage’, and any idea that companies would respond better to the former than the latter is rather implausible.

The content of the duty is not specified in any detail. But let’s explore the tree-planting example. Suppose policy required every adult to plant at least one tree a year. That could help in mitigating climate change (though it is pertinent to ask how much else would be required to maintain the habitability of the planet).  But it could waste resources to little effect in places where the climate or soil is unsuitable for tree growth.  It could also do harm where tree-planting conflicts with other important land uses, or where the uptake of water by trees within a river basin would lead to reduced water flows with adverse downstream effects on agriculture or wetlands.  So a sensible policy focused on tree-planting would have to be more complicated, perhaps permitting people in places unsuitable for trees to enter into offsetting arrangements with those in more suitable places.

One way to give content to the duty while addressing such issues would be a policy telling each person in what way they should contribute to public environmental goods and positive externalities, but with the contributions tailored to people’s local circumstances: tree-planting in region A; rainwater harvesting in B; fish restocking in C; and so on.  At the other extreme, another would be to allow people discretion as to the nature of their contributions, but to require each person to submit an annual statement of their contribution for review by the authorities.  Intermediate arrangements would probably be more workable, for example, a list of acceptable contributions from which people could choose any one, or a system of points for contributions with an annual points target for each person.

What emerges from this is that giving practical form to a duty to create public goods and positive externalities would not meet Laitos’ requirement that policy should be simple.  Any apparent simplicity is entirely due to Laitos’ abstract presentation of his ideas.  Putting it into practice would raise the same sorts of issues as existing environmental policies, such as whether policy-makers have adequate technical advice, conflicts between different environmental objectives, and political feasibility.

Then there is the issue of enforcement.  Laitos states that environmental policy needs a “stick” instead of “carrots”, contrasting the “carrots” of subsidies with the “stick” of his positive duty (p 200).  This is an odd comparison for two reasons.  Firstly, subsidies are only one of many instruments used by existing environmental policies, and as a means of incentivising firms to reduce pollution are generally considered by economists to be inferior to taxes (because subsidies can also encourage new firms to enter polluting industries).  Secondly, a duty is not itself a “stick”. A “stick” would be a penalty that an authority could impose for failing to fulfil a duty.  One might have expected here a discussion of the process for imposition: how evidence might be gathered; whether the process would be administrative or judicial; types of penalty; and so on.  As throughout the positive parts of his book, however, Laitos seems uninterested in such details.

In summary, therefore, although Laitos is right to highlight the failure of many existing environmental policies, the alternative he offers owes whatever plausibility it may have almost entirely to its abstract formulation.  As soon as one attempts to flesh it out with practical detail, it becomes apparent that it would encounter much the same complexities and difficulties as existing policies.

Notes and References

  1. Laitos J, with Okulski J (2017) Why Environmental Policies Fail  Cambridge University Press.  All page references above are to this book.
  2. Wikipedia: Montreal Protocol – Effect https://en.wikipedia.org/wiki/Montreal_Protocol#Effect
  3. US Environmental Protection Agency: Lead and Copper Rule – A Quick Reference Guide https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=60001N8P.txt
  4. Wikipedia: Flint Water Crisis https://en.wikipedia.org/wiki/Flint_water_crisis
  5. Wikipedia: Nuclear Power in Germany – Closures and Phase-Out https://en.wikipedia.org/wiki/Nuclear_power_in_Germany#Closures_and_phase-out
  6. Der Spiegel (22/11/2017) Can Germany Break its Lignite Habit? http://www.spiegel.de/international/business/energy-transition-blocked-by-brown-coal-a-1179537.html
  7. UN Framework Convention on Climate Change – Kyoto Protocol – Targets for the first Commitment Period http://unfccc.int/kyoto_protocol/items/3145.php
  8. My policy in this blog when setting out the views of others is normally to paraphrase and rarely to quote. This is partly because a paraphrase highlighting key points can be more concise than a quotation, and partly for copyright reasons. I hope that my paraphrases are fair and, as ever, would appreciate advice of any misrepresentation.
  9. Wikipedia: Atmosphere of Earth – Evolution of Earth’s Atmosphere https://en.wikipedia.org/wiki/Atmosphere_of_Earth#Evolution_of_Earth.27s_atmosphere
  10. Wikipedia: Last Glacial Period https://en.wikipedia.org/wiki/Last_glacial_period
  11. Carey B, Live Science (20/7/2006) Sahara Desert was Once Lush and Populated https://www.livescience.com/4180-sahara-desert-lush-populated.html
  12. The government regulations on the 5p charge are at http://www.legislation.gov.uk/uksi/2015/776/contents/made
Posted in Environment (general) | Tagged , , , , , , | Leave a comment

Improving on GDP

The winning entries for the Indigo Prize highlight both the importance of improving economic statistics and some of the challenges of doing so.

The launch of the Indigo Prize by LetterOne is a welcome initiative which recognises both the limitations of conventional economic statistics and the need for a little razzmatazz to bring the issue to the attention of a wider public (encouragingly there have been reports in several UK newspapers).  Entrants were asked to submit a 5,000 word article presenting a design for a new economic measure for global economies and explaining how it should be used to improve the measurement of GDP.  The new measure was to take full account of social and economic factors and the impact of creativity, entrepreneurship and digital skills.

The remit made no mention of environmental or natural resource issues.  A clue as to why may perhaps be found in the contribution on the Indigo website by Mikhail Fridman, a co-founder of LetterOne, in which he rejects the view that there is an imminent problem of depleting resources.  Fortunately, both entrants and judges appear to have taken a broad view of their task.

Given the current state of knowledge, no entry could have been expected to offer a definitive solution to the problem of economic measurement.  Nor is it the sort of problem that lends itself to such a solution: what is best will depend on economic circumstances and (so far as high level presentation is concerned) on the capacity of policy-makers and the public to interpret economic indicators.  Nevertheless, both of the joint winning entries are insightful and thought-provoking (and contain much more than I shall mention here).  That by Haskel et al begins by identifying what it regards as two key strengths of GDP, namely, that it avoids double-counting of intermediate outputs, and that it adds outputs of different goods in an objective way using market prices as weights. Recognising however the limitations of GDP, it proceeds to identify various ways in which GDP could be “fixed” and “extended” to address those limitations while preserving its strengths. It outlines how it might be possible to define and estimate a single measure of well-being that would be a substantial improvement on conventional GDP.

The entry by Coyle and Mitra-Kahn proposes, for the long term, an approach which differs from that of Haskel et al and of conventional GDP in two key ways.  One is that it focuses on assets rather than annual flows.  If such an approach were adopted, people would no doubt consider annual increases or reductions in assets as well as their absolute magnitude. But that itself is quite different from focusing on annual GDP, much of which (eg most consumption) does not contribute to the maintenance and growth of assets.  The second difference is that it rejects the idea that economic debate should focus on a single key number, proposing instead a dashboard presenting separate numbers for broad asset groups.

Of these differences it is the first that is the more radical.  It has never been the case that public debate on economic matters focused solely on GDP.  A fairly conventional ‘dashboard’, though perhaps not presented as such, would include measures of output, employment, inflation, poverty and perhaps a few more variables.  The interesting question here is how far it is sensible or meaningful to aggregate items using weights, with the implication that categories that cannot sensibly be aggregated should be treated as separate indicators within a dashboard.  But this is really two questions, one presentational and one technical.  For presentation to the general public, and as a high-level framework for public debate, a dashboard of six indicators (the number proposed by Coyle and Mitra-Kahn) seems about right, in the sense that it would inform but not overload with detail the lay person with no special interest in economic policy.  The technical question is how many indicators we need to include relevant items that cannot meaningfully be aggregated using weights, and the answer to that question is probably many more than six (within natural capital, for example, consider whether it is meaningful to aggregate petroleum, fresh water, forests and fisheries).

So what do the winning entries say about environmental and resource issues?  As part of ‘fixing’ GDP, Haskel et al propose treating enhancement of the natural environment (eg planting a forest) as investment (an addition to GDP) and degradation (eg destruction of a coral reef) as disinvestment (a deduction).  Presumably they would treat depletion of minerals in a similar manner.  This approach requires valuation of the enhancement and degradation, which is challenging both in principle and in practice.  Taking the coral reef example, an issue of principle is whether the loss of value attributed to destruction of a given quantity of reef should be constant, or depend on how much coral remains elsewhere in a country’s territorial waters, or elsewhere in the world (on the basis that scarcity of a good  increases the marginal value per unit of what is left).  A practical issue is the collection and assessment of data from below the ocean surface.

A more fundamental problem is how the measure obtained by adjusting GDP is to be interpreted. Conventional GDP is, at least, a fairly good measure of economic activity (though it could be improved even in that respect).  The measure that would be obtained after the above adjustments would be such that a year-on-year reduction could reflect either a fall in economic activity or a reduction in net enhancement of the natural environment.  Most people, I imagine, would want to know which, rather than focusing on a measure which combines both.

As part of ‘extending’ GDP, Haskel et al propose that it should count future as well as current consumption (a line of thought which may suggest a concern with resource depletion, although it is introduced in the context of the balance between consumption and investment).  Their reference to Weitzman (1) indicates that what they have in mind here is that suitable adjustments can convert GDP to a concept of net national product (NNP) which measures what Weitzman calls “the stationary equivalent of future consumption”.  In my view this is not a workable or especially useful proposal.  It isn’t workable because the adjustments needed to obtain Weitzman’s NNP depend upon shadow prices of capital goods, and these shadow prices are obtained by solving a long term dynamic optimisation problem requiring assumptions about production technology far into the future (2).  It isn’t very useful because, although a future-oriented NNP concept may sound as if it has something to do with sustainability, “the stationary equivalent of future consumption” is not the same as sustainable consumption.  The relation between the two concepts is quite complex, as explained in this post, and if (implausibly) we had the data needed to calculate the former, we would be able to calculate sustainable consumption directly.

Within the dashboard approach of Coyle and Mitra-Kahn, natural capital is one of the six groups of assets.  Their definition of natural capital as “the renewable resources provided by nature” (p 6) should presumably have read “renewable and non-renewable”, since their chart showing natural capital declining in England Wales (p 12) includes minerals, oil and gas.  They state that measurement of natural capital is currently very incomplete and based on market prices.  The former is correct, but the latter seems to refer to official statistics, and overlooks the considerable academic literature on non-market valuation techniques including the hedonic pricing method (for local environmental quality) and the travel cost method (for recreational sites).

I would agree however that, in the context of a high-level dashboard of six items, it is appropriate that one of the six should relate to the environment including natural resources, and should be a capital measure.  A slight disadvantage of such a measure is that it will tend to ignore the short-term (though often recurrent) effects of non-cumulative pollutants.  Greatly outweighing that is the fact that most of the more important environmental and natural resource issues (eg depletion of minerals, deforestation, over-fishing, greenhouse gas emissions) involve long-term effects that a suitable capital measure should reflect.

Coyle and Mitra-Kahn note that valuing their asset groups, including natural capital, is a major challenge.  I wonder however if it is necessary or appropriate to value each group.  If each item in a dashboard is quantified in monetary terms, that could be taken as an invitation to add the values together and focus on the total rather than the separate items.  Having distinct indicators that cannot be aggregated would seem more consistent with a dashboard approach (think of a car dashboard including speed, fuel level and oil pressure, each in separate units that could not be aggregated, even if bizarrely someone wished to do so).  In the case of natural capital, there is also what might be termed the ‘baseline problem’.  For example, should the UK place a positive value on the fact that it does not suffer from serious earthquakes?  Or should the valuation baseline be defined to include ‘absence of serious earthquakes’ so that the value of natural capital for Japan, say, would include a negative element for earthquake risk?   There is a case therefore for including natural capital in terms of some non-monetary measure based on an assessment of its adequacy to sustain current living standards into the future, although so far as minerals are concerned that leads back to the issue of assumptions about future production technology.  So although the idea may have merit, working out the detail would be just as challenging as trying to value the whole of natural capital.

Notes and References

  1. Weitzman M L (1976) On the Welfare Significance of National Product in a Dynamic Economy  The Quarterly Journal of Economics  90(1) pp 156-162
  2. Weitzman, as above: the “investment prices” he introduces on p 158, which vary over time, would have to be obtained as shadow prices in the solution of the dynamic optimisation problem set out on p 159. The relevance of future production technology is implicit in Equation (6) on p 158 which features the marginal product of each capital good.
Posted in Environment (general), Macroeconomic Theory | Tagged , , , , , , , | Leave a comment

A Difficulty in Assessing Sustainability

Long-term forecasting of the aggregate production function is essential for assessing sustainability, but very difficult.

Is our current living standard sustainable?  Or does depletion of non-renewable natural resources such as fossil fuels and metal ores mean that it must eventually decline?  And if the latter, what lower standard of living, if any, would be sustainable?  A well-founded sustainability criterion, enabling us to give clear and convincing answers to those questions, could make an important contribution to public debate on economic policy.  In particular, it could help counter the view that growth in GDP is the key indicator of economic performance.

Qualitatively, it is fairly clear how it might be possible to sustain consumption while non-renewable resources are progressively depleted.  We start from two reasonable assumptions:  that production technology will permit a considerable degree of substitution between inputs, and that there must always be at least some input of non-renewable resources. Inputs of non-renewable resources must therefore be ever-decreasing so that reserves are never totally exhausted (in mathematical terms, the stock of non-renewable resources must be asymptotic to zero).

To maintain output and consumption in those circumstances, there must be increasing inputs of reproducible capital (which includes both capital equipment and infrastructure and renewable natural resources).  Furthermore, it will not be sufficient simply to maintain output: output must grow to provide not only for consumption but also for the necessary investment in reproducible capital and for offsetting depreciation of the increasing stock of that capital.

Whether this scenario is feasible will depend on a number of variables: the initial stocks of capital and resources, resource extraction costs, production technology, population and labour inputs, current consumption and (if we are considering the economy of one country rather than of the whole world) opportunities for trade. There are difficulties of measurement and/or forecasting associated with several of these variables. This post will argue that, in particular, the forecasting of production technology presents a major challenge in assessing sustainability.

That might seem a surprising claim.  After all, economists have put much effort into the estimation of aggregate production functions and into economic forecasting using models in which a production function is one component (1).  The problem is that, to assess long-term sustainability, we need to estimate the relation of output to inputs for future periods in which the ratio of reproducible capital to input of non-renewable resources will be vastly greater than at present.  And that is very difficult.

To illustrate the point, I shall consider the assessment of sustainability within a model having the following simplifying assumptions:

  1. Output of a single good which can be either consumed or invested as reproducible capital.
  2. A Cobb-Douglas production function with technical progress Y = AKαRβ(1.01t) , where K is man-made capital, R is use of a non-renewable resource S, extracted at nil cost, t is time in years from the present, and A, α and β are parameters.
  3. Constant population and labour (this is why, although production requires labour, labour does not appear as a variable in the production function).
  4. Depreciation of reproducible capital at 5% per annum.
  5. Initial stocks: 100 units of K and 100 units of S (the respective units need not be the same).

The assumption of technical progress at a constant 1% per annum, implicit in 2 above, is made to facilitate a focus on the rest of the production function and keep the mathematics of the model reasonably simple.  The actual rate of future technical progress is difficult to forecast, but that is a reinforcement, not a criticism, of the argument presented here.

The question I shall explore is whether, within this model, annual consumption of 100 units (the same units as those of K) is sustainable.  This will depend on the parameters of the production function.  Consider the following functions:

F1:  Y = 15K0.3R0.2(1.01t)

F2: Y = 24K0.3R0.1(1.01t)

This pair of functions has the following property: if K approximates to 100 and K/R to one, then they yield very similar values of Y (because 15 x (1000.1) ≈ 24).  This is illustrated, for the case t = 0, by Chart 1 below.

Suppose we had to decide whether the production function is F1 or F2 on the basis of input and output data for recent periods in which K/R is in the vicinity of one.  Given the inevitable random variation in output due to other variables, this would be extremely challenging. Statistical tests would at best point to one function as slightly more likely than the other.

At the very much higher K/R ratios that sustainability requires, however, the difference between these functions becomes very significant, as shown in Chart 2 below (note the log scale on the horizontal axis).

Because the same inputs yield more output with F2 than with F1 whenever K/R exceeds one, we expect that sustainable consumption will be greater given F2. To find out how much difference this makes I set up the model in discrete form as a spreadsheet with one row per year for 5,000 years.  Key features of the spreadsheet are as below:

  • Consumption in each year equals that for year 0.
  • Growth of capital equals output less consumption less depreciation.
  • Marginal product of capital is calculated directly using the standard formula (for a Cobb-Douglas production function) MPK = αY/K.
  • Marginal product of the resource is calculated directly (MPR = βY/R) for year 0, but subsequently using the Hotelling rule according to which the rate of growth of the marginal product of the resource equals the marginal product of capital (2).
  • Extraction and use of the resource after year 0 is calculated backwards from its marginal product and output (to avoid circularity, the output figure used is that for the previous year).
  • Resource stock after year 0 is that for the previous year less extraction and use for the previous year.
  • Consumption and extraction / use of the resource for year 0 are trial values. A pair of trial values is considered feasible if it generates time paths for the variables in which, during years 0 to 5,000, the resource is not exhausted and output is never less than consumption.

The maximum trial value of consumption consistent with feasibility as defined above is an approximation to maximum sustainable consumption (it’s only an approximation because of the discrete spreadsheet approach and because of the 5,000 year time horizon).  It was found that:

  • With production function F1, constant consumption of 100 units is unsustainable as maximum sustainable consumption is approximately 79.7 units.
  • With production function F2, constant consumption of 100 units is sustainable as maximum sustainable consumption is approximately 117.6 units.

So the assessment of whether consumption of 100 units is sustainable rests on the weak foundation of whether F1 or F2 in Chart 1 above provide a better statistical fit to current and recent data.

A possible objection is that the implicit sustainability criterion, that is, the feasibility of consumption at 100 units per annum forever, or for 5,000 years, is too demanding.  Suppose instead, therefore, that we set the time horizon at just 100 years.  Surprisingly, perhaps, this makes very little difference to the conclusion.  For F1, maximum sustainable consumption increases from 79.7 to 80.0 units.  For F2, it increases by such a tiny amount that on rounding to one decimal place, maximum sustainable consumption is still 117.6 units.

Addendum 1 November 2017

A further possible objection is that if the initial stock of the resource is 100 units, then it is not very plausible to assume that resource use in the preceding years used to estimate the production function was in the region of 100 units per annum.  To address this, I considered the effects of assuming initial stocks of the resource of 1,000 units and 10,000 units (in each case with initial capital of 100 units and over 5,000 years).  With 1,000 units, maximum sustainable consumption was 138 units with F1 and 151 units with F2.  With 10,000 units, maximum sustainable consumption was found to be higher with F1, 239 units, against 194 units with F2.  The important point, however, is that for each of these variations, maximum sustainable consumption depends significantly upon the production function.

The spreadsheet used to derive the above charts and results may be downloaded here:

Constant Consumption with Different Production Functions Adam Bailey

Notes and References

  1. One example is the NIESR’s NiGEM model: see https://nimodel.niesr.ac.uk/nigem-intro/nigemintro.php?t=1&b=1&w=1
  2. A derivation of this form of the Hotelling rule may be found in Perman R, Ma Y, McGilvray J & Common M (3rd edn 2003) Natural Resource and Environmental Economics Pearson Addison Wesley  pp 660-1, the rule being equation 19.42j.
Posted in Macroeconomic Theory | Tagged , , , , , , | Leave a comment

A Valuation Case Study: the Great Barrier Reef

A valuation of the Great Barrier Reef illustrates many common issues in the economics of environmental valuation.

A recent report by Deloitte Access Economics (1) estimates the asset value of the Great Barrier Reef at A$56 billion (2).  This is the sum of:

  • Value to local recreational users and tourists from elsewhere in Australia: A$32 billion, estimated by the travel cost method (pp 39 & 40).
  • Value to Australians of the Reef’s existence, regardless of having visited it or intending to do so; A$24 billion, estimated by contingent valuation (p 34).

The main data source was a representative survey of over 1,000 Australians (p 30). The survey also included 500 residents of 10 other countries world-wide.

The report also identifies the following components of value for which it does not offer numerical estimates:

  • Value to tourists from outside Australia: not estimated due to data limitations (p 85).
  • Existence value to non-Australians: not estimated as difficult to allow for contextual cultural factors, language barriers and purchasing power differences in using survey responses from some countries (p 35).
  • The value of certain ecosystem services provided by the Reef such as maintenance of water quality and storm protection for the adjacent coast (pp 40 & 42): not estimated as difficult to separate from ecosystem services provided by neighbouring terrestrial and river ecosystems (p 84).
  • “Traditional owner value” to peoples who have lived in the vicinity of the Reef since before Australia was settled by Europeans, relating to cultural heritage, sacred sites and archaeological sites: not estimated because such unique sites lack the substitutability on which non-market valuation relies (pp 45-7).
  • Brand value, that is, the contribution of the Reef to “Brand Australia”. This is given a whole chapter of the report (pp 50-7) which is not easy to summarise.  One key point is how far the contribution is from the existence of the Reef and how far from perceptions of Australia’s performance as guardian of the Reef (p 56).

Taking the above together the report suggests, reasonably, that the full value of the Reef is much more than A$56 billion.  Some, though not all, of the points below tend to reinforce that conclusion.

There are other components of value which the report does not mention.  One is the possible medicinal use of plants and animals found in the Reef (3).  A negative component – illustrating the general point that natural capital is often associated with natural hazards  –  is its role as a shipping obstacle and hazard (4).

As is common with environmental valuation studies, therefore, the components of total economic value fall into three categories: those that can be measured with the available data; those that are in principle measurable but for which the necessary data is difficult to obtain; and those that are intrinsically unmeasurable (but still perhaps important).

As a contingent valuation sceptic (see this post), I would have been inclined to put the existence value of the Reef in the unmeasurable category.  A specific difficulty with the approach adopted is that respondents were asked how much they would be willing to pay weekly to “guarantee” that the Reef is “protected”, having been told that such a payment would be in the context that all Australians would have to pay (p 80). This is a strange question given that global climate change is a major threat to the Reef.  It seems quite possible that some respondents viewed the payments as to guarantee that the Australian government would do what it could to protect the Reef (eg from risks originating within Australia such as mining activity), while others viewed them as guaranteeing the preservation of the Reef, implying protection against all risks including climate change.

The remainder of this post considers the method used to estimate the value of the Reef to tourists from within Australia. The individual version of the travel cost method was used, given that individual data was available and showed good variation between individuals in visit numbers (p 78), so that it could be expected to give more precise results than the alternative zonal method (5).  The variation between individuals was not just due to luck: it was facilitated by the decision to ask respondents how many times they had visited the Reef in the last five years (not merely the last year).

With the individual method, the first step is to regress individual visit numbers against individual travel costs to the Reef and other variables that may influence visit numbers.  The report does not state the full regression model, but from the survey questions (pp 76-82) it appears that the other variables included age, gender and education.  Variables that seem not to have been included are income and travel costs to one or more substitute sites.  Although income is mentioned in discussion of the method (p 85), there is no indication that income data was collected.  This may have been because it was anticipated that such data would be difficult to obtain (US studies often collect income data but it may be that willingness to disclose one’s income varies between cultures).  The issue of substitute sites may have been ignored because of the complexity of allowing for many possible substitutes, and lack of consensus in the literature as to how this should be handled.  Whatever the reasons, omission of these variables can result in omitted variable bias leading to a biased estimate of the travel cost coefficient and hence of the site value, although the magnitude and direction of bias will depend on the circumstances (6).

Mention of substitute sites may seem surprising given the unique nature of the Great Barrier Reef.  But a substitute, in economics, does not have to be a perfect substitute.  One definition is that two goods are substitutes if their cross-elasticities of demand are positive.  Applied to two tourist sites, that would be the case if a higher travel cost to either were associated with a higher demand for visits to the other – a plausible scenario.

The decision effectively to ignore travel time (p 85) is surprising. Common practice is to include the value of time as well as expenditure on transport and accommodation within travel cost to the study site.  The report cites three Australian studies which applied a zero value to travel time, but this is rather selective even among Australian studies (7), and certainly not representative of the global literature including US and UK studies.  Admittedly there is no consensus on how the value of time should be determined.  But to assume a nil value is likely to result (other things being equal) in under-estimation of the value of the Reef.  A fairly simple and still conservative alternative would have been to value time at a suitably small fraction (say a quarter) of the average hourly wage, and to include the effects of different fractions in the sensitivity analysis.

The statistical methods used to estimate the trip-generating function and then derive the value (consumer surplus) per visit (A$662) are not described in full, but appear from the outline provided (pp 85-6) to have been appropriate.  As is common in individual travel cost studies, negative binomial regression was used because of overdispersion (p 86), which is a slightly misleading term for a feature (not a defect) often found in the distribution of individual visit numbers, namely that the variance is greater than the mean.

I would have liked to see some consideration of what potential visitors might have done instead if the Reef had not existed.  Many would probably have visited other tourist sites, increasing the value of those sites and offsetting to some degree the lost value associated with the Reef.  A case can be made that the most useful concept of value for a recreational site is not the gross site value but the contribution which the site makes to the total value of all such sites (8).

To obtain the total annual value to tourists, value per visit was multiplied by the annual number of visits to the Reef, which was estimated as 2.3 million (p 86).  This number was inferred from Tourism Research Australia (TRA) data on numbers of visits within Australia to regions adjacent to the Reef.  However, these numbers include visits to the regions but not to the Reef itself, and some crude round-number assumptions, claimed to be conservative, were made to eliminate such visits (p 86).  An alternative approach would have been to include in the survey a question designed to distinguish visits to the Reef itself from visits to the adjacent regions, and to use that data to estimate, for all visitors to those regions, the proportion who visit the Reef itself.

The final step was to capitalise the stream of annual values so as to obtain the asset value.  The result is heavily influenced by the choice of time horizon and discount rate, as is illustrated by the sensitivity analysis (p 88).  A time horizon of 33 years was adopted, one stated reason being the severe threats to the future health of the Reef (pp 87-8).  Within that time frame, annual consumer surpluses as calculated from the survey data were discounted at a rate of 3.7%, determined using the Ramsey formula (p 87):

Social discount rate  =  Rate of time preference + [Annual growth rate of consumption

                        x Minus the elasticity of marginal utility with respect to consumption]

Given assumptions of a very low rate of time preference (0.05%) and an elasticity of 1, the discount rate largely reflects the growth rate of consumption, which was assumed to equal the average GDP growth rate over the previous 30 years (which can be inferred to have been 3.65%).

While the particular figures used in the capitalisation could be challenged, there are some more fundamental issues here.  One could value the Reef on the assumption that it will degrade along a defined path, or on the basis of its remaining in its current condition.  The key question here is not which of these scenarios is more likely, but which basis of valuation will yield more useful information.  If what we are interested in is the value which is potentially at risk from degradation of the Reef, then it is valuation on  the ‘current condition’ basis which is more useful, and the ‘degradation’ argument for the 33-year time horizon does not apply.

Whether future economic growth is likely to be at a similar rate to that of the previous 30 years can be debated.  My instinct would be to calculate the central estimate of value on a zero-growth basis, and to consider the effect of positive rates within the sensitivity analysis.  If, however, a positive growth rate is assumed, then for consistency it needs to be considered that, since expenditure on tourism is discretionary, demand for tourism is likely to be highly income-elastic.  Subject to the condition of the Reef,  and to the adequacy of infrastructure to accommodate visitors, demand for visits might be expected to grow even faster than GDP.  The report, however, does not consider whether the number of visits to the Reef may change in future.  The implicit assumption is that annual visits and annual consumer surpluses will remain as estimated (9).

Notice that the combination of a very low rate of time preference and a zero rate of economic growth will yield, via the Ramsey formula, a very low discount rate.  In conjunction with a long time horizon this could imply an asset value many times higher than the report’s A$56 billion.

Notes and References

  1. Deloitte Access Economics (2017) At what price? The economic, social and icon value of the Great Barrier Reef  https://www2.deloitte.com/content/dam/Deloitte/au/Documents/Economics/deloitte-au-economics-great-barrier-reef-230617.pdf   All page references are to this report.
  2. A$ = Australian dollars. A$56 billion is equivalent to GB£33 billion or US$43 billion.
  3. Bruckner A (2002) Life-Saving Products from Coral Reefs Issues in Science and Technology XVIII(3) http://issues.org/18-3/p_bruckner/
  4. See for example The Guardian (6/4/2010) The Great Barrier Reef scandal https://www.theguardian.com/world/2010/apr/06/great-barrier-reef-ship-aground
  5. King D M. & Mazzotta M Ecosystem Valuation – Options for Applying the Travel Cost Method  http://www.ecosystemvaluation.org/travel_costs.htm#OPTIONS
  6. Re omission of substitute site variables see Caulkins P, Bishop R & Bouwes N (1985) Omitted Cross-Price Variable Biases in the Linear Travel Cost Model: Correcting Common Misperceptions Land Economics 61(2) pp 182-7
  7. Two Australian studies which applied positive values to travel time are: a) Lansdell N & Gangadharan (2003) Comparing Travel Cost Models and the Precision of their Consumer Surplus Estimates: Albert Park and Maroondah Reservoir Australian Economic Papers 42(4) p 403  b) Whitten S & Bennett J (2001) A travel cost study of duck shooting in the Upper South East of South Australia  Private and Social Values of Wetlands Research Reports No. 7  https://crawford.anu.edu.au/pdf/staff/jeff_bennett/wtlndrr07.pdf
  8. This point is briefly noted in Bateman I (1993) Valuation of the environment, methods and techniques: revealed preference methods, in Turner R (ed) Sustainable Environmental Economics and Management: Principles and Practice Belhaven Press, London  pp 218-9
  9. This is shown by the fact that a stream of consumer surpluses of A$662 x 2.3 million = A$1.523 billion annually for 33 years, discounted at 3.7% per annum, approximates to the A$29 billion stated on p 39.
Posted in Environment (general) | Tagged , , , , , , , , | Leave a comment