How much can we nudge?

In our new paper, my colleagues and I show how defaults affect voluntary investments into climate action even when they come at considerable price tags. Overall, defaults have a considerable impact on behavior, but become less effective when prices rise.

Like Comment
Read the paper

Shortly before making the transition into my Ph.D. program, I was handed a book by my soon-to-be Ph.D. advisor. My advisor asked me to ditch my original seminar topic and focus on a novel behavioral science approach called “Nudge”. Richard Thaler and Cass Sunstein had just published the first edition of what would become their global bestseller and I was intrigued by the promise of behavioral science as part of the solution to all kinds of policy-challenges. Years later, my own research has focused on green energy nudges and we could show in a large-scale randomized controlled trial that green energy defaults led to a dramatic, tenfold increase in the uptake of green energy contracts in Germany.

In our new paper, just published in Nature Human Behaviour, we sought to investigate how much the effectiveness of such default nudges may depend on the cost they impose to a decision maker. Powerful nudges – such as green energy defaults – often only come at a small price tag for consumers. For example, in our own trial, the additional cost of green energy was negligible, about ten Euros on average per household and year. Other nudges – such as defaults that affect saving rates for pension scheme – lead arguably to financial improvements for consumers. Finally, defaults that change organ donation status-quos do not have an impact on the living self.

In our new paper, my colleagues and I were interested what happens to the effectiveness of defaults when they do not come at negligible price tags or even benefit the decision maker? Prior to our paper, the effectiveness of nudges in such “high cost” scenarios was largely unknown. Cass Sunstein put out the question in an editorial in Nature Human Behaviour, stating  that “[…] a great deal remains to be learned. It makes sense to think that when and where the cost of green energy is high, the opt-out rate will be higher, but how much higher?” In essence, this question was the key motivation of our paper.

We were lucky as we could test this hypothesis in a field setting with actual and consequential behavior, enabled through a collaboration with a platform that allows compensation of flight-related emissions. The platform is owned by a large European airline. In our data, we could analyze more than 30,000 trips compensated by roughly 11,000 passengers. We collected our data just before the COVID pandemic started. Offsetting cost varied from negligible amounts (0.51 EUR) to extremely high stakes (2,877.17 EUR), allowing a systematic analysis of the default efficacy under varying financial consequences.

The decision context was the following. Customers who consider compensating their flight are directed to the platform. They can make the decision “how fast” to offset their flight. Fast, in this case, means purchasing “Sustainable Aviation Fuel” (SAF), which is an alternative jet fuel that reduces emissions by 80% compared to standard kerosene. Each monetary amount invested into SAF causes a change in the airline’s fuel mix. Slow, in our case, meant a compensation of flight-related emissions through reforestation, realizing the emission reduction over twenty years. Debates about the efficacy of such offsets aside, this context allows us to test the effectiveness of climate action nudges with similar environmental impact under different cost scenarios.

On the platform, customers can make a choice of any combination between SAF and reforestation. The default nudge was implemented through a slider that stopped at pre-defined places. Sometimes the slider would stop at 8 years, thereby compensating relatively fast. Other times, the slider stopped at 18 years, thereby compensating relatively slow. Because flight origins and destinations were different, booking classes varied, and the slider stopped at various places, we could observe how the default affects decision-making in comparison to the cheapest offset available (i.e. only reforestation).

Overall, we found that a striking 43% share of decision makers accepted the default. This includes many decision makers for which the default came at considerable cost. The highest accepted default was a whopping 776 Euros. And among those customers who left the default by adjusting the slider, the default continued to influence their decision making. Perhaps more importantly, we found support for what Cass Sunstein had suspected. If cost go up, the drop-out rate becomes higher. We observe neatly that the cost has a profound impact on whether a default is accepted or not. Finally, if people deviate from the default, they do so in a way that favors their narrow monetary self-interest. Deviators from the default overwhelmingly choose an offset that is slower and thereby cheaper for the individual.

What can we learn from our paper? First, our results seem to suggest that nudging even works at considerable individual cost. Second, the fact that people respond sensitively to the price tag provides reason to belief that we cannot infinitely nudge decision makers into accepting outcomes against their fundamental preferences. Finally, if we carefully re-interpret the evidence to “nudging works better when price incentives align better (i.e., less bad)”, our paper may provide hope for nudging energy decisions in Europe’s current crisis. If decision makers can actually make money, defaults targeted at climate action (e.g., lowering thermostats in the winter to save energy, increasing refrigerator temperature) may help to reduce energy demand. But of course, this is best tested in a next study.


Sebastian Berger

Assistant Professor, University of Bern