To reset organizations for the long run, we need to more visibly connect how our day-to-day work activities contribute to climate change. One good approach to this is internal carbon pricing. Used by Microsoft and others, it involves a company charging its own units for their carbon emissions linked to activities ranging from heating their offices to purchasing supplies.

To better understand how this works in practice, I spoke with Casey Pickett, director of the Carbon Charge at Yale, one of the first universities to levy a carbon charge internally and a model that other organizations are looking at. Here is our conversation, lightly edited for clarity:

What’s the story behind internal carbon pricing at Yale?

The story behind Yale’s carbon charge is that in 2014 [Nobel-Prize-winning economist] Bill Nordhaus was on a panel for Earth Day. When asked, what could Yale do to be more of a climate leader or something like that, he said, well, we could charge ourselves for the carbon. I gather that elicited polite titters in the audience. But one graduate student doing a joint degree in the School of the Environment and School of Management picked up on it and ran with the idea, and got some other students interested. They wrote a white paper for Dan Esty’s class on environmental management. And when they submitted it he said, this is really good and you should actually send it to the president, which they did. And the president and the provost really liked the idea of an internal carbon fee. President Salovey asked Bill Nordhaus whether he would chair a task force on that idea, which that student then staffed. Her name is Jen Milikowsky.

That task force baked the idea for six months, thought through how it might work. And then that student and another recent graduate staff member went forward and, with a steering committee overseeing it, implemented a pilot study trying out four different approaches to internal carbon pricing on five buildings each. It was a group of 20 buildings they were trying different things on, and we ultimately decided to go with the revenue-neutral approach that we’re using. That approach is most similar to the carbon fee and dividend model that Citizens Climate Lobby is pushing for. The proposal from the Climate Leadership Coalition follows that same model.

How does it work?

The provost office charges the different planning units at Yale. Planning units are our top-level organizations. We’ve got about 50 planning units. These are schools, departments like athletics, different offices. About half of that 50 have financial responsibility for one or more buildings. There’s plenty of planning units that occupy but don’t have financial responsibility for buildings. So we charge the planning units that have financial responsibility for buildings a fee for their carbon emissions. And at the same time we give them a return. So everybody gets both a charge and a return. And what they actually are responsible for is the difference between those two numbers. The way that we calculate the charge is quite simple.

We take energy meter data from natural gas, electricity, steam, and chilled water, convert that into emissions in the form of metric tons of carbon dioxide emissions, and then charge everybody $40 per ton of emissions that they’re responsible for. Now that then aggregates to a total amount of carbon charge across the university, several million dollars. And we give that all back. So the provost’s office, the university center, retains none of that money. It’s not revenue. We give it all back. The way that we give back is based on people’s baseline emissions. So we’ve looked at how much emissions each building was responsible for in the baseline period, which we’ve set as 2011 to 2015.

There are 267 buildings that participate in the carbon charge. We’ve got almost 600 buildings, but those buildings account for about 75% of our building-related emissions. The rest are small or the ownership structure is confusing, or I literally can’t figure out who to send an energy report to. Let’s picture that all of Yale emits a certain amount this year. Let’s say it’s 5% less than it was for 267 buildings in the baseline. So you think of the graph that has the baseline number, and then the current year number is a lot lower than it.

Then picture three individual building scenarios. Building A has the exact same percentage change as the others—it reduced its emissions by 5%. That means it’s going to actually end up paying and getting back nothing, because it’s charged number and it’s returned number are identical. So the net is zero. Building B increases its emissions while Yale had decreased them. So it’s going to get a greater charge than it did a return. It’s going to end up having to pay because it’s a positive difference. Building C decreases its emissions by more than Yale did. It’s going to end up getting a greater return than it’s charged. It’s kind of making out financially.

Your focus is exclusively on energy consumption and not on things like faculty travel emissions or purchasing at this point?

That’s right.

How could this model work for a business?

There’s a couple of things that you need to make an approach to internal carbon pricing like this one work. There are several different approaches that you can use. The proxy price is the obvious other kind of approach. To make a fee-based approach work, you need different business units to play the game. You can’t play monopoly against yourself. And then you need the data. You need good enough data that people have enough faith in this. But if you have those things, you should be able to do it. It takes some figuring, but it’s certainly doable.

Are there any important lessons you’ve learned in implementing this?

The most important lesson is don’t let the perfect be the enemy of the good. The system that we have is certainly not perfect. And the analogy I like to use is it’s like a sport. We don’t expect every basketball team that plays to have gotten the same amount of sleep last night, or traveled the same distance, or to have the same amount of money to support their players and attract players. It’s about getting the rules fair enough that everyone’s willing to play the game. There will be inherent unfairnesses. And I’m not sure we’re quite there yet. In our system, honestly, we’ve got some more work to do to get it more fair. The biggest issue is around the baselines and how do you calculate those baselines? Buildings are built at different times, which that’s all fine—our formula handles that. Our formula is forward-looking, so it deals in percent changes, not absolute changes. Every building is on the same footing and larger, less efficient buildings aren’t at a disadvantage. In a sense, more efficient buildings may be at a disadvantage because they might have less low-lying fruit. Because it’s all about your percent change. Some of the critiques we get internally are yes, I’m doing a big renovation. And we’re expanding and that’s going to raise my carbon fees—but I was asked to do those changes by the university center. How is that fair? Or, we’ve got a new building. What are you going to give me for a baseline? That’s a tough one. So those are some challenges.

Another thing to really think through is do you have a system in which you’ve created an incentive and people are able to respond to it. There’s a lot more flexibility that even not wealthy people have in their consumption patterns that people who are managing buildings that are in a big company or institution might not have. So that that’s the thorniest issue we’ve got to deal with.

What do you do in the case of a new building?

We give them the best estimate from our facilities department about what their energy consumption is going to be.

What has the reception been?

I did 42 different meetings around campus from one-on-ones with deans to 150-person meetings with entire staffs. And what I got from that was literally three people were brave enough—I was coming from the provost office so I’m sure I was sheltered from some of the feedback and negative feedback that people were feeling potentially—but what I got were three people brave enough to say some version of they don’t think that’s a good idea. We really shouldn’t do that, I disagree with the very concept. I brought two of those folks on board. This was in the fall of 2016. My first day in the office was the Friday before the Tuesday of the 2016 election. What I heard was I think a lot of reaction to that election in a way. People were looking for an opportunity to do something. I had maybe about a dozen people say some version of I’m so glad we’re doing this, this is the right thing that you’re doing, I’m really enthusiastic. And then the vast majority of what I heard in each of those meetings was almost word for word—it was a little crazy actually—it was okay, I get it. I’m on board. It’s going to be a pain, but I’m willing to go for it. .

All those meetings I was having were in the spring of 2017. And then it went live July 1, 2017. Since then, one of the challenge we’re having is people are oddly quiet. I get some feedback from a small handful of folks. A group has emerged to provide constructive critique. I’m working with them really closely on developing what might the next part be, how could we tweak this? And that’s going really well. We’ve got some ideas.

Do you have plans to expand it?

We’d love to. We don’t have the data yet. What we need to be able to do that is the most obvious candidate is expanding to air travel. Only 20% of Yale’s air travel is booked centrally. We’ve got a data collection problem there. Purchasing is the next obvious place, but to figure out the carbon impact of the 50 most carbon intensive things that Yale buys, even if it were the 10 most intensive things Yale buys, that’d be pretty incomplete and relies on life-cycle assessment work, which is very difficult.

What has been the impact on energy consumption?

We don’t know. It’s hard to collect that data. We’ve been collecting data, but we intentionally chose at the beginning not to create a control group. We wanted to roll this out as widely across campus as we could. There are several buildings that we didn’t include, but we didn’t include them with bias. So we can’t do a randomized controlled trial. I’ve been working on other research methods, other statistical analytical methods. The next most promising is called synthetic control, where you create a control group out of peer data. I’ve been working really hard to try and get peer data. We have 18 institutions that have contributed some data, but only five of the data sets are sufficiently complete that they make a good group. And that ends up not being a big enough group to make the formula work for that analytical method.

What has been the aggregate energy consumption change?

It’s been down about 5% since compared to the baseline period. But the question is what’s cracked that. We have a very active group at Yale that is doing work, not driven by the carbon charge, just doing work independently on energy reduction.

What is the magnitude of the charges and rebates for one of the units?

There’s the range of returns and charges that goes right through the numbers zero. But there are lots of units that pay in the $1,000 to $10,000 range. There are a few units that get over $10,000 in returns and there are a few units that get between $10,000 and $100,000. There’s one unit that usually ends up with well over $100,000 in net charges.

But we shouldn’t think of the net here. This is a really difficult to communicate point. And this is one of the communication challenges that I’ve had actually is helping people see that they actually shouldn’t be looking at the net because they should be thinking like an economist and then looking at the marginal costs.

Think of it like you would if your employer sent you on a trip and give you $75 a day to spend on food, and didn’t ask you to bring back receipts, the per diem approach. Buy a fancy $76 vegan dinner and say, this is so cool. I only spent a dollar on this meal. But that’s not the right way to think about it. You can think about it as the employer put $75 in my pocket. And every dollar that I spend comes out of my pocket. So my incentive is actually to go buy a burrito for five bucks and pocket the $70. That’s how people should be thinking about this. You’re going to get the return no matter what the charge. So if you’re charged ends up being $2 million on a $20 million utility budget. Even though your return is about the same amount, you should be focused on the $2 million. It’s a hard message to get across.

You said one unit has over $100,000 in a net charge.

It’s a unit that has been asked by the university center to grow organically. In the Yale as world analogy, that unit is China. Our charge system works pretty well when units are relatively stable and they’re changing things in the margins. If you’re growing like crazy, it does feel like a perverse incentive in a way. On the other hand, the biggest impact we think this will make is as a governor on growth.

So their operating budgets weren’t increased and the money is just part of the department operating budget and they didn’t get additional funds to compensate for that?

Correct. They do not get additional funds and they have to figure out a cover any charge within their existing budget. Any new money that comes back, they can do whatever they want with.

So why do this, given that you’re not sure what the impact is?

That’s why because it’s an experiment we can run. For first couple years of this project, I was working with the World Economic Forum and World Bank to do a webinar series where we interviewed companies about their internal carbon pricing. We did like seven decent webinars. But we were scraping the bottom of barrel to get people to talk about this. We don’t have the same challenges that a company has. I joke that I’m always willing to show people my innermost spreadsheets, and we can tell people how this is going on and we can be honest about it. That’s a service that we can provide to the world. And as a research university it’s useful.