Monthly Archives: March 2011

183 – Diminishing marginal benefits of environmental projects

Environmental managers face the question of how large in scale to make their environmental projects. An important consideration is whether the projects would face diminishing marginal benefits as project scale increased.

This is a sequel to PD182, which covered increasing marginal costs of environmental projects. “Diminishing marginal benefits” is another fundamental concept in economics – one that is relevant to both production and consumption.

On the production side, the benefits from using more of a production input usually decrease as you use more and more of it. A classic example is the application of fertilizer to crops. The almost universal pattern is as shown in Figure 1 – yield increases steeply at first, but as more fertilizer is added, the rate of yield increase falls towards zero.

Figure 1. Diminishing marginal benefits from applying fertilizer to a crop.

On the consumption side, the pleasure that people get from consuming a unit of a good usually diminishes as they go to higher and higher levels of consumption. Personally, I don’t consume much milk, but the bit I do have (in my tea and on my breakfast cereal) is pretty important to me. However, if I were to increase my milk consumption beyond that, it would give me very little additional pleasure, if any. This pattern of falling additional pleasure from increased consumption is almost universal across all types of goods, not just food.

My question here is, as we devote more and more resources to generating environmental benefits of a particular type in a particular area (e.g. more agricultural land is managed to benefit the environment in the Loddon River catchment of Victoria), are we likely to see diminishing marginal environmental benefits – a flattening out of the benefits like in Figure 1?

Perhaps not in every case, but the evidence indicates that there are many cases where this occurs, either in production or consumption of environmental goods, or both.

Like the fertilizer example, it may be that a change in land management is initially highly effective in generating environmental benefits, but the gain in benefits falls as you increase the area of land that has been changed. For example, I was involved in some research in Germany that looked at the economics of riparian vegetation buffer strips of different widths (Seiber et al., 2010). We found a pattern very similar to Figure 1. It was estimated that buffer strips 3 metres wide would reduce pesticide movement into rivers by 61%, while 30-metre strips would reduce pesticides by 94%. Increasing the area devoted to buffer strips by a factor of 10 increased the environmental benefits by a factor of only 1.5. Increasing the width further to 50 metres made almost no further difference to the pesticide load in the river – a striking case of diminishing marginal environmental benefits. It’s not unusual for models of land management for other types of problems to show a broadly similar pattern, although rarely that extreme.

Like the milk example, people’s pleasure or happiness as a result of “consuming” particular environmental improvements often tends to flatten out. There is a wealth of evidence on this. Economists’ studies of intangible environmental values often find that people’s valuations of larger scale projects are not much higher than for small projects, implying that there are rapidly diminishing marginal benefits. For example, Bateman et al. (2005) could not find a significant difference in total values between protecting 4 and 400 lakes. Other studies do find benefits in going from small-scale to large-scale projects, but the additional benefits are often pretty small. To some extent, it may be that there are limitations in the research methods used in these studies, but even allowing for that, the evidence for diminishing marginal benefits is powerful.

Diminishing marginal benefits are the opposite of what tends to be assumed by the proponents of “landscape-scale change”. They seem to assume that, by fixing the whole landscape, you can get exceptional aggregate benefits, even if the benefits from fixing up parts of the catchment would be modest. Perhaps there are scale thresholds that need to be crossed before large benefits can be generated, resulting in increasing marginal benefits. Could that be true? I suppose it could for certain types of issues in certain situations.

The only clear evidence I’ve ever seen for increasing marginal environmental benefits on the production side was from a modelling study of revegetation to reduce dryland salinity in the Western Australian wheatbelt (George et al., 2001). This showed that there might be (slightly) increasing marginal benefits in some catchments, but not others. Thresholds are often talked about, but I’ve never seen clear evidence of them generating increasing marginal benefits. (If you know of some, I’d be keen to see it.) Given the clear evidence for “diminishing” in at least some cases, it wouldn’t be safe to just assume “increasing” without some evidence. (For some issues, the relationship is approximately linear – neither increasing nor diminishing.)

On the consumption side, I’m aware of one study that found increasing marginal benefits for the beneficiaries of a particular environmental project (Holmes et al., 2004). But this is a rare exception to the usual results.

Where we face diminishing marginal environmental benefits, the implications for environmental managers are the same as for increasing marginal costs. They mean that the greatest environmental benefits in aggregate can be achieved through a larger number of well-targeted, modest-scale projects, rather than a small number of huge projects. They are another reason why an enthusiasm for “landscape-scale change” is likely to be ill-conceived in many cases, especially given that environmental budgets are small relative to the levels needed for effective landscape-scale projects. Diminishing marginal benefits and/or increasing marginal costs mean that support for landscape-scale projects would reduce environmental benefits overall by drawing resources away from projects with higher marginal benefits.

On the other hand, this is not an argument for funding a multitude of tiny projects (the sort of “vegemite” approach for which the Natural Heritage Trust was criticised). Next week I’ll pull together these last two Pannell Discussions and suggest what they do mean for environmental programs.

David Pannell, The University of Western Australia


Bateman, I.J., Cooper, P., Georgiou, S., Navrud, S., Poe, G.L., Ready, R.C., Riera, P., Ryan, M. and Vossler, C.A. (2005). Economic valuation of policies for managing acidity in remote mountain lakes: Examining validity through scope sensitivity testing, Aquatic Sciences 67, 274-291.

George, R.J., Clarke, C.J. and Hatton, T. (2001). Computer modelled groundwater response to recharge management for dryland salinity in Western Australia, Advances in Environmental Monitoring and Modelling 2, 3-35.

Holmes, T.P., Bergstrom, J.C., Huszar, E., Kask, S.B. and Orr F. III (2004). Contingent valuation, net marginal benefits, and the scale of riparian ecosystem restoration, Ecological Economics 49, 19-30.

Sieber, S., Pannell, D.J., Müller, K., Holm-Müller, K., Kreins, P. and Gutsche, V. (2010). Modelling pesticide risk: A marginal cost-benefit analysis of an environmental buffer-zone programme, Land Use Policy 27, 653-661.

182 – Increasing marginal costs of environmental projects

One of the key insights of economics is the idea of increasing marginal costs – as you produce more and more of a good, production usually gets more expensive per unit. This idea usually relates to manufactured goods but is highly relevant to environmental ‘goods’ as well. It has implications for environmental management that often seem to go unrecognised.

Increasing marginal costs is one of the most fundamental ideas in economics. It is the first thing you learn about when studying the economics of production. To force more output out of a given production process, you may have to pay workers more in overtime, or pay more for inputs, or use a more expensive production system, or reduce production of something else that was making you money, so the cost of production goes up.

Evidence for this is everywhere, if you look. For example, consider the fact that, when the price of wheat goes up, farmers around the world quickly increase their production of wheat. This reflects that farmers need a higher wheat price to cover the higher costs of more production. (In the long run, there can sometimes be economies of scale, which go the other way, but that’s another story.)

For some years, a lot of my research has focused on environmental issues related to land use and land management. For issues like biodiversity protection, dryland salinity, soil erosion and nutrient pollution in waterways, the key to improving environmental outcomes is often to change land use or land management.

You could, if you had economic tendencies, think about this as a process of producing or “supplying” environmental goods. The goods are more biodiversity or less-polluted rivers, and the production process consists of changing land use or land management. The more you change, the more environmental goods you produce.

Should one expect increasing marginal costs in this environmental “production” process? Absolutely.

If only very small changes were required, you could cherry pick the cheapest options for change. By picking the cheapest options, you could achieve the greatest environmental improvement that was possible within the environmental budget. As you move to steadily larger targets for change, you aren’t able to be so choosey any more, and are forced into steadily more expensive changes.

For example, if the key response to an environmental issue is conversion of crops to native vegetation, you’d tend to start with those areas where crop production was least profitable (where the “opportunity cost” was lowest). As the required changes increased in extent or intensity, you would have to convert steadily more-profitable crop land, so that, at the margin, the cost of providing greater environmental benefits would increase. The same sort of thing would apply if the changes required were reductions in fertilizer use, or changes in grazing management. In order to maximise environmental outcomes, you’d start with the cheapest options and work up.

Figure 1 is a real example. It is taken from a study we did calculating the cost of achieving different reductions in flows of phosphorus into the Gippsland Lakes in Victoria. Because more ambitious targets for nutrient reductions require us to take up less and less attractive management options, the total cost goes up at an increasing rate. It’s actually quite cheap to reduce nutrients by 10%, but its very expensive to reduce them by 40% – much more than four times as much.


Figure 1.

Now, in the world of land and water conservation, an idea that has a lot of currency is that of seeking “landscape-scale” change. The hope is that, by fixing up environmental problems across the whole landscape, you’ll generate benefits that are more than the sum of the parts. It is assumed that by fixing the whole landscape, you can get exceptional aggregate benefits, even if the benefits from fixing up parts of the catchment would be modest.

That may or may not be true in particular cases (see PD183), but either way, it is not just benefits that matter. Especially when considering management of large areas, it is highly likely that making changes across the whole landscape will result in exceptional aggregate costs. For landscape-scale interventions to be better than more targeted interventions, the escalation of benefits as the area treated increases would have to be greater than the escalation of costs. Whether that’s the case would need to be assessed on a case-by-case basis.

Thus, even if one is confident that there will be exceptional aggregate benefits from a landscape-scale response, one shouldn’t just assume that this is the way to go without considering the cost side of the equation. When the environmental budget is limited (and it always is), it may be that the greatest environmental benefits in aggregate can be achieved through a larger number of more modest-scale projects, rather than a small number of huge projects.

David Pannell, The University of Western Australia

181 – Answering questions after presenting a seminar or conference paper

You’re well prepared for a seminar or conference presentation, but as the day approaches, doubts and fears emerge. You worry about messing up in question time. What if they find an error in your work? What if they ask hard questions, or dumb questions, or make long rambling speeches instead of asking a question? Here are some ideas about how to handle these and other issues in question time.

Doing a good job in question time is mostly a matter of common sense, but it can help to have thought in advance about how you might handle tricky situations.

In general, it’s best to answer the question that is actually asked, if you can, rather than a vaguely related question (but there are exceptions, as outlined below). It’s worth taking a bit of care to think about what the questioner is really driving at. Ask for clarification if necessary.

If you have understood the question, and really don’t know the answer, don’t try to bluff it. Say that you don’t know. This looks much better to the audience. Attempts to bluff will be obvious to at least some, and perhaps all, of the audience.

It is fine to say something like, “That’s a good question. I’ll need to give it some more thought afterwards. Let’s talk later.” And move on to the next questioner.

An alternative is to say that you don’t know the answer to that question, but it reminds you of this other question, and give the answer to the other question. This violates my first suggestion, but you can get away with it if you admit you don’t know the answer to the real question.

It can be tempting to make up an answer to a question that you actually don’t know the answer to. This can be OK, as long as you start by saying that you don’t really know the answer, but here are some thoughts. It’s not a good idea to talk as if your response is definitely the answer, because a perceptive audience member could make you look silly.

Don’t be defensive about your work when you should actually concede a weakness that someone has pointed out. (A hard thing to do!) By all means point out any mitigating circumstances or rationalisations for you having adopted an approach with a weakness, but don’t deny that it is a weakness. Every study has weaknesses and limitations of one sort or another, and its just good scientific practice to acknowledge yours.

Sometimes you might give an answer, but then later in question time realise that your answer was wrong. In my view, you actually look good if you publicly admit your earlier error, and provide the correct answer. It shows you have high integrity and are really concerned to answer accurately.

It is fine to take a little while to think about a question before answering. Say something like, “let me think about that for a moment”, and give yourself 5-10 seconds.

Unless the questioner is dragging things on and wasting time, don’t start answering the question until they finish asking it. This gives you a bit of extra time to think, and makes sure you hear the actual question.

If you don’t understand the question (e.g. due to a strong accent, poor audio conditions, or poor wording of the question), ask the questioner to repeat it or to express is differently. If after one repetition you still don’t understand it, it’s probably best to offer to talk to them one-to-one later on, after question time has finished.

Having a co-author in the audience can be helpful if you get stuck, but guard against directing most questions to them. If you are a co-author in the audience, make sure you stay alert through the talk and the questions, because you might be called on at any moment. Co-authors should not offer criticisms of the work or the talk during question time. If criticisms are necessary, save them for later. Co-authors should be sparing in answering questions. Leave it to the speaker, if possible. But do help them out if they need it. A co-author who does respond to a question should make it clear to the audience that he or she is a co-author.

Sometimes audience members make a statement, rather than asking a question. You can respond to or expand on the statement if you want to, but you also have the option of saying, “I’ll take that as a statement, rather than a question” and moving on to the next questioner.

If a questioner is hogging the floor, the chair of the session should cut them off, but failing that, you can do so if you wish. If they haven’t asked a question yet, ask them what there question is. If they keep dragging out or repeating the same question, tell them that you’ve understood the question, and proceed to address it, if you can.

In a large room or one with poor acoustics, it often helps if you repeat the question for the benefit of those in the audience who couldn’t hear it. (This is assuming that there is no roving microphone for questioners to use.) The same strategy can be helpful if the question is ambiguous. This at least makes it clear which question you are actually answering.

If someone asks more than one question at a time, write down a couple of words about each one to remind yourself of what they are as you’re working through them. (You need to have a pen and paper with you for this.)

If the question relates to a particular slide in your PowerPoint presentation, go to that slide before you start answering. Here’s a nice PowerPoint trick to get you to the slide quickly. With the slide show still running, type in the slide number (it doesn’t display anywhere) and press enter. Alternatively, press ctrl-S and you get a menu of the slides. These commands are particularly useful if you have custom animations that you want to skip over, or if you have a lot of slides and you want one in the middle.

How long should your answer be? Not too long, not too short, just right. Thirty seconds to a minute is about the right length of answer for many questions. However, if you’ve provided a good answer in 10 seconds, don’t feel that you have to keep elaborating and extending it. Short answers are good.

It may be possible to anticipate the questions you will be asked. If so, be prepared. Prepare an answer in advance, so you are convincing and clear in your response. I often include additional slides at the end of my PowerPoint file (after the final slide of my talk) just in case I get asked certain questions. They don’t get used every time, but it’s good to have them there for when they are needed.

Preparing questions will, hopefully, reduce the risk that you’ll get rattled during question time. Try to stay calm and measured in your responses, no matter what happens. Take your time if you need to.

What if you find yourself in exactly the situation you’ve been dreading, and a questioner finds a serious error in your analysis? I haven’t had that when I was speaking, but I’ve been in the room when one of my co-authors was speaking and facing this challenge, so I was partly in the hot seat too. I could see he needed rescuing, so I jumped in and said it was an excellent point, and we would have to revise the model and re-do the analysis. This is the only thing you can do really. Definitely don’t bluff. It’s not as bad in reality as you might expect. The audience tends to accept that mistakes happen. They are likely to keep asking further questions, despite the error. Self-deprecating humour can be effective in this situation, if you can manage it.

What if someone claims to find an important error in the middle of your talk? If they are right, it may be that the rest of your results are wrong. If you are sure that you can deal with the issue convincingly without taking up much time, do so. In all other situations, my suggested strategy is to thank them for their advice, and suggest that you re-visit the issue during question time. Usually, seminars and conference presentations are on a tight time schedule, and you don’t want to lose half your time in an argument. Even if they are right, it is probably best not to just abandon your prepared talk (except in extreme cases). Rather, have a discussion in question time about the implications of their point.

What if you get a really aggressive questioner? After one talk to a large conference audience, I had a guy (a university academic) absolutely let rip at me with abuse and contempt. He was passionate, to say the least. In this situation, you’ve got one big thing on your side: most of the audience feels almost as uncomfortable as you do. When I smiled and thanked Dr Angry for his question, quite a few people laughed, which helped me a lot. Fortunately, I knew he was talking rubbish, and I was able to refute his points in what I hoped was a cool, calm way. I was trying to pretend I was cool and calm, anyway. It certainly would not have been a good idea for me to respond aggressively – the calmer the better. Even if he had been right in his comments, the same advice holds – try not to look rattled, and respond calmly. Maybe a good strategy would be to say, “I’d be happy to talk about it afterwards, if we can do so civilly”. That makes you look reasonable, and includes a bit of a put-down of the questioner (which you might find satisfying).

One of the things I find most difficult to deal with is when someone asks an absolutely ridiculous question or one that reflects a strong ideological position or some whacky new-age belief (more common in a public forum rather than an academic or professional meeting, although you never know!). The question deserves a response something like “You idiot!!”, but you have to respond respectfully. Sometimes a decent response would require much too long, because you would have to start by addressing the foundations of their misguided beliefs. Sometimes you just don’t know where to begin. How should one respond? Remember there is no way you’ll change their beliefs. Really, your aim should be to get off that topic as quickly as possible and onto the next question. Maybe say you don’t know the answer. Maybe say that’s an interesting question that you’ll need to give more thought to. Maybe answer a different question. Maybe say you’ll have to agree to differ about the assumption that underpins their question. Definitely don’t ask them to talk to you afterwards!

Sometimes an apparently dumb question reveals that you actually haven’t explained something properly. Assume that the fault is yours (poor communication) rather than theirs (poor understanding). Try to re-state and clarify the point of misunderstanding. Sometimes this requires some interaction with the questioner to identify the exact point of confusion.

Finally, a risk to guard against is that you might relax at the end of your main presentation, and give a flippant answer to a question, such that you open up a can of worms. Remember that the presentation doesn’t end until the questions are over.

David Pannell, The University of Western Australia


Thanks to Michael Burton for the idea for this PD, and to Michael, Ben White and Fiona Gibson for ideas for content.

Further Reading

Pannell, D.J. (2009). PowerPoint Tips, Pannell Discussion No. 147, here.