Somewhere near the end of the booking flow, after you have chosen your seat and declined the extra legroom, a small prompt appears. Would you like to offset the carbon emissions from your flight? The cost is usually a few euros. Most people scroll past it. The airlines know this, which is why several Nordic carriers have started folding the offset into the ticket price instead, making it automatic, invisible, one less thing for the passenger to think about. The design logic is intuitive: if you want people to do the right thing, make it the default. Our research suggests the logic is wrong.
Together with Steven Schoenmaker and Maja Fors, I ran an experiment to test whether different carbon offsetting designs change how people travel afterward. We recruited 150 Nordic consumers through Prolific and assigned them randomly to one of three conditions. In the first, participants booked a flight with no mention of carbon offsets. In the second, the offset was integrated into the ticket price, visible but automatic. In the third, participants were offered the choice to add a voluntary offset for five euros.
All three groups booked the same outbound flight, Stockholm to Oslo. The question was what they would do next: for the return trip, they could choose between flying back or taking the train.
The results were clear in direction, if modest in scale. Participants who had been offered the voluntary offset were 2.4 times more likely to choose the train. Participants whose offset had been integrated into the ticket behaved no differently from participants who had seen no offset at all. The effect was strongest among the subset who actually chose to pay the five euros: they were 4.6 times more likely to take the train home.
This is counterintuitive. The design logic behind integrated offsets assumes that making sustainability automatic, removing the friction, produces better outcomes. Our findings suggest the opposite. The friction was the point.
The theoretical framework we started with was moral licensing and cleansing. Most people have experienced moral licensing without knowing the term: you recycle diligently, so the long-haul flight feels earned. You bought the organic milk, so the plastic bag does not register. The good act creates a surplus that the less good act spends. Moral cleansing runs in the other direction: people who feel they have fallen short of their own standards seek out compensatory behavior. The psychology literature has found both patterns across a range of sustainability contexts, from eco-labels to energy conservation campaigns.
We hypothesized that voluntary offsetting, because it requires an active choice with a real cost, would lower people’s moral self-image more than the passive, integrated version. That dip in self-image would then motivate cleansing behavior, choosing the train. The mechanism made theoretical sense.
It did not hold up empirically. Both offsetting conditions lowered moral self-image compared to the control group, and by roughly the same amount. There was no difference between the voluntary and integrated groups, and no difference between people who chose to offset and those who declined. The moral self-image pathway was a dead end.
But the behavioral difference remained. Something about the voluntary condition changed what people did next, even though it did not change how they felt about themselves in the way we expected.
We think the explanation may lie in agency rather than guilt. Voluntary offsetting puts the passenger in a position of active decision-making about the environmental consequences of their flight. The integrated version does not. When the offset is folded into the ticket, it is something that happens to you, a line item you may or may not notice. When it is offered as a choice, you have to reckon with it. You have to decide whether the environmental cost of your flight is something you are willing to pay five euros to address. That reckoning, even if you decline it, appears to carry forward into the next decision.
This aligns with research showing that perceived agency increases concern with making sustainable choices. When consumers feel their individual decision matters, they behave differently than when the outcome feels predetermined. The voluntary offset may function less as a guilt mechanism and more as a prompt: a moment in the booking flow where the environmental dimension of the trip becomes something you personally have to respond to, rather than something the airline has already handled on your behalf.
The implications are uncomfortable for the airline industry, which has largely moved toward integrated models. Integrated offsetting has real advantages: it eliminates free-rider behavior, it guarantees funding for offset projects, and it does not require passengers to opt in. SAS, for instance, automatically offsets all youth, staff, and member tickets. From an emissions accounting perspective, this is straightforwardly better than relying on the small fraction of passengers who would voluntarily add five euros to their booking.
But our findings suggest that integrated offsetting may also neutralize something. When the airline handles the environmental cost for you, the passenger’s relationship to that cost becomes passive. The offset exists, but it does not prompt anything. It is background noise in the booking confirmation, a line item that requires no response and generates none. The behavioral spillover, the possibility that engaging with the environmental dimension of one trip might change the next one, disappears. The airline has solved the problem for you. Which means you no longer have to solve it yourself.
This does not mean airlines should abandon integrated programs. Carbon offsetting itself remains deeply contested. The temporal and spatial distortions of compensation schemes, the difficulty of verifying that offset projects deliver what they promise, the risk that offsets function as permission structures rather than transitional measures: these criticisms are serious and unresolved. Our study does not address whether any particular offset project effectively reduces emissions. What it addresses is something narrower but still consequential: whether the design of the offsetting interaction changes subsequent behavior.
The answer appears to be yes, with caveats. Our sample was small, Nordic, and responding to a hypothetical scenario rather than booking an actual trip. Nordic consumers may already be more environmentally conscious than travelers in other markets, and a hypothetical choice is not the same as standing in Oslo Central Station deciding whether to spend six hours on a train. The effect sizes were modest. We conducted a single experiment and have not replicated it. These are real limitations, and we name them as such.
But the pattern is worth paying attention to, because it cuts against a widespread assumption in sustainability design. The instinct to remove friction, to make the sustainable choice the default, to automate good behavior so that individuals do not have to think about it, is well-intentioned and often effective. Default organ donation, automatic enrollment in retirement savings, pre-selected green energy tariffs: these are genuine successes of choice architecture.
Our finding suggests that in some contexts, the thinking is the intervention. When you remove the decision, you also remove the moment where a person connects their behavior to its consequences. And that connection, uncomfortable and effortful as it is, may be what changes what happens next.
The chapter this post draws from was co-authored with Steven Schoenmaker and Maja Fors and is forthcoming from Routledge.