Does It Work To Pay People Not To Cut the Forest?

It might seem only fair, then, that if we want people to leave their wildland intact for the greater good, we should pitch in to alleviate the opportunity cost: the money a person might have made from developing the land instead. But how do we organise this in a fair and effective way?

Economist Sven Wunder, principal scientist at the European Forest Institute and senior associate at the Center for International Forestry Research, has spent years investigating just this. Wunder, who shared his insights in two 2020 articles in the Annual Review of Resource Economics, told Knowable Magazine that evidence for the effectiveness of such projects is often sorely lacking — but it’s within our reach to do better.

This conversation has been edited for length and clarity.

What is the philosophy behind payments for environmental services (PES) schemes, and what is your own experience with them?

For a long time, the conservation movement mostly held the idea that conservation is for everybody’s good, including landowners, and that we just need to give people environmental education to persuade them. But in terms of what they might otherwise have done with the land, it’s not always for the owners’ own good not to develop it, at least not without compensation.

I’m specialised in incentive schemes for the conservation and sustainable use of forests, where a landowner gets paid or otherwise rewarded if they agree not to cut down the forest, by selling certain land rights for a certain period of time. This typically restricts their use of their own private land. I’ve worked in the tropics for more than 20 years, including in Brazil, Peru, Ecuador, Colombia and Indonesia. I’m still doing so now, but my current research also includes Europe.

I should add that in the European setting, impact evaluations have not been common at all. When we’re doing something in the Global South, everyone wants to know if we are getting value for money for our investments. But when it comes to our own backyard, we’re often more reluctant to get the hard facts, because some subsidies may work better for sustaining the income of farmers, for example, and maybe not so well for the environment, and we don’t want to dig too deep to find out.

Are there studies comparing the effectiveness of, say, education versus compensation?

This is what the articles I coauthored for the Annual Review of Resource Economics are about; they are taking stock of the available evidence. There are some educational programs in there — we call them “enabling interventions.” But there aren’t many studies of those, and the record is not great.

Ideally, how should we go about evaluating the impact of various approaches?

We take inspiration from how different treatments are compared in medical science, by comparing an area that receives a certain “treatment” to a similar control area that doesn’t. The gold standard is often argued to be a randomised controlled trial — areas or landowners are randomly assigned to the treatment or control group. But it’s often hard to convince organisations to roll things out in that way. Sometimes they can be persuaded to develop a program in which some people receive benefits now, and others only later on. But even that is often difficult to do, because the ones who have to wait will likely not be happy with the arrangement.

Studies of this kind are therefore still relatively rare, but an exemplary one was published by Seema Jayachandran and colleagues in Science in 2017. They studied the effects of payments for forest conservation in 121 Ugandan villages, 60 of which were randomly selected to participate in the program. Using satellite imagery, this study showed convincingly that tree cover loss near those villages was only half of that observed around the other villages — 4.2% on average, instead of 9.1%.

Using a solid methodology to show such a clear “treatment” effect from payments to farmers, that study had quite some impact. And after the project ended, the World Bank did a follow-up study to see what would happen when the money stopped coming, which they published as a report.

In a way, the World Bank results were also comforting: Deforestation did pick up again to the same speed as before, but that didn’t eliminate the loss that had been prevented — total tree cover lost near the project villages did not catch up with that around other villages. So when you stop paying, you haven’t eliminated the problem, but you have at least mitigated it for the duration of the project. This is important: In the context of climate change, even the temporary gains count.

Do you think it would be important for all projects of this kind to test their approach first in this way, by initially implementing their program only in some of the targeted areas, and comparing them with the areas that haven’t received an intervention yet?

Yes — and if a randomised trial is not possible, other impact evaluation methods may work.

One example is the Trans-Amazon project in the Brazilian Amazon, where a nonrandomly selected group of farmers was receiving payments for the environmental services of keeping their forest intact. The others received no payments, only technical assistance.

Early on, we surveyed those farmers at the same time as two control groups of farmers: one within the project that received only technical assistance, but no payments, and another one outside the project that received no treatment at all, but was a good match in basic characteristics such as farm size, income, distance to roads and markets and so on.

Thanks to impact evaluations, we could see that farmers with payments had effectively reduced their deforestation compared with the matching controls, so we could attribute desirable conservation outcomes to a specific policy.

In other cases where the intervention has already been implemented, you obviously cannot survey people’s behaviour in real time. You will have to rely on preexisting data such as satellite images or agricultural censuses.

It is all about credibly simulating what would have happened without the targeted intervention.

I imagine some project managers may not be quite sure how to properly go about this. What would be your advice to them?

I think it is ideal for researchers to work directly with implementers and assist them in adapting the intervention along the way, to increase its likely impact. Researchers can also help to communicate the findings more widely, so that other projects can learn from them how to do better.

Many implementers may be a bit afraid of researchers, because obviously, there is a risk that it will turn out that what they have been doing did not achieve what it was supposed to. In that sense, I think it’s the donors that would really need to press for these impact evaluations to be an integral part of implementation, especially for larger programs that are spending a lot of money.

Why is it necessary to compare a project area with other areas? Why isn’t it enough if people see clear progress – say, reduced deforestation – in the area they work in?

Because you need to account for what would have happened in the absence of your intervention. You need to assess all the things beyond the treatment that may also be affecting the forest. For instance, some projects that are selling carbon credits for the deforestation they claim to have prevented are in the middle of nowhere, where deforestation was very unlikely anyway.

We have recently finished a global evaluation that was published in Science in August. In this study, we’ve looked at projects certified by Verra. This is the organisation that developed the leading standard for voluntary offsets markets — which allow companies and individuals to buy carbon credits to compensate for some of their own greenhouse gas emissions. The data are publicly available.

We’ve found that more than 90% of these credits were, at the time of our evaluation, non-performing. This means they haven’t actually prevented any deforestation, and therefore did not have a positive climate impact. The main reason they didn’t perform is that the baseline deforestation rates they assumed for the area were much too high, which resulted in grossly overestimated benefits.

So would you argue, then, that payments-for-environmental-services schemes should be employed only in places where deforestation poses a big risk?

In my opinion, our conservation actions generally need to be targeted more to threat, yes. For instance, we often see that people opt to only enroll the parts of their forestland they will not be clearing anyway, in which case the payment-for-environmental services schemes won’t help. For instance, in the Peruvian Programa Nacional de Conservación de Bosques, a pilot program, many people, when asked, chose to preserve the high and far areas that couldn’t be converted to agriculture. Such a “high and far” selection bias is also quite common when countries or organisations decide which areas to protect, going for the low-hanging fruits.

Of course, even if we protected all the areas currently under threat, there is still a risk people might instead go to areas that are unthreatened today — what we call leakage. What we find, however, is that leakage remains partial, and is largest for small projects, because in this setting, people can more easily move pressures “next door,” so to speak.

For larger projects, some high-value activities like oil palm plantations may still go somewhere else, but for things like slash-and-burn agriculture, moving may not be worth it.

One risk of focusing all efforts in places where the risk of deforestation is highest is that failure may be more likely there, in which case you haven’t achieved anything.

That’s true, and that’s why you may want to focus on sizeable but manageable threats, where a compensation payment might tip the balance and really make a difference. Also, you don’t want to pay only the bad guys: What about areas with Indigenous people who are managing their forest well — should they not be rewarded? I think rewarding the good guys, in high-forest, low-deforestation countries like the Guyanas in South America or Gabon in Africa, where pressures are currently low but growing, is an important complementary approach.

But the strategic focus should be on high-risk areas.

Another thing you warn about in one of the papers, however, is that projects tend to attract participants who are already oriented towards conservation. For maximum impact, there would probably be more rejoicing in heaven for a handful of sinners who convert.

Exactly. In scientific terms, this is called “adverse selection bias”: The good guys will be first in the queue to line up for the money. In certification schemes to label sustainable products, for example, it’s extreme — because there typically are no mechanisms to focus only on the most vulnerable areas or the most dangerous actors. The standard is the standard, so all people who are already doing things the right way are entirely entitled to sign up and get paid.

But if, conversely, you pay only the bad guys, your intervention will not be perceived as socially acceptable, and that might kill the whole project. So you have to strike the right balance.

Would this be the main reason then that certification schemes for sustainably managed forests or supposedly forest-friendly agricultural products are not effective? I think it’s one of the worst-performing approaches in your analysis, right?

We’re still lacking a larger body of evidence, but I do think certification has this particular problem, that it’s more exposed to adverse selection bias. It also has the problem that it often depends on a premium to be paid by consumers, and often, those premiums have not been that large. There are also problems of different certification standards out there that might be confusing to consumers. So you would need a certifier of certifiers — which brings us back to the problems faced by Verra.

A fundamental criticism of payments for environmental services is that once we start paying certain people for not doing something we don’t want them to do, the ones who weren’t doing it in the first place may come to expect money as well. How can we deal with that?

This is what we call “crowding out intrinsic motivation.” There is a nice collection of case studies about this in the journal Ecological Economics, and the findings are quite mixed.

Crowding out is found in some cases, but not as often as some political scientists would predict. We mostly find that payments leave people’s underlying motivations intact. And in some situations, we even see “crowding in,” where the fact that some people get paid makes them and others realise that they’re doing something good for society, and so they keep doing it even after the payment scheme ends.

Obviously, when conservation is working well out of intrinsic motivation, there is no reason to go in and start paying people there. I think some projects that caused “crowding out” have probably applied the instrument in the wrong setting. To me, that is another argument for payments and carbon credit schemes to focus on areas with documented threats — and leave other places alone.

There is also some risk of a “magnet effect,” which means that payments attract migration into the area. It’s a potential unintended effect that we need to be careful about, and I think it’s a reason for not making incentives too large. It should be worthwhile for people to participate in these things, but there shouldn’t be huge gains, because that may have unforeseen consequences. You want to pay enough to tip the balance and make sure local people are better off, but you shouldn’t overdo it.

One of the reviews you coauthored concludes that there is “shockingly little evidence” about the impacts of payments for environmental services, and that “the little available is troublingly disappointing.” People who read that might wonder why we should keep doing this at all.

I think it’s good for us to face these results, because there are some illusions about what can be achieved. In many ways, we are still at the beginning of the road. A lot of the debate is about the money — that our current issues can be resolved with this or that many billions to fill the financing gap. That is obviously in the interest of the organisations that push this, because it would boost their budgets.

But I think it ignores the fact that we often know too little about how to actually use that money most effectively. That’s still kind of scary for me, how little acknowledgement there is of the fact that we don’t design our interventions well enough to be able to show value for money in the end.

That said, there are also many schemes out there that are working well, and I’m certainly not saying we shouldn’t invest in these things anymore — I’m saying we should do it better. The biodiversity and climate crises are here, and we need to do something about it. And as our review also points out, while payments for environmental services are no magic bullet, they still do better than most other conservation approaches.

Of course, some people will say we shouldn’t rely so much on offset-type solutions and that we need to solve the root problems instead. For the climate crisis, I think that may be true, but for the loss of biodiversity, the big struggles — such as land-use change, deforestation and illegal hunting — are mostly in the tropics, where many threatened species live.

I realise that if you take our conclusions about conservation incentives in isolation, they may sound very pessimistic. But I’m more optimistic myself — because of the good examples that are out there. To me, the important part is that people who are putting money into this should think more about how to find out what works and what doesn’t.

We’re often not patient enough. We tend to move from one conservation fad to the next. We try something new, we get disappointed, and so we move on to something else, and we are once again overly optimistic about what we could achieve. What we should be doing is take more time to identify the issues and refine a promising approach until we make it work.