Extrinsic rewards and crowding-out of prosocial behaviour

Can offering monetary rewards reduce prosocial behaviours like recycling? We find that rewards can both increase and decrease prosocial behaviours, and that the existence and size of the crowding-out effect depends on reward size and pre-existing levels of prosocial motivation.

Like Comment
Read the paper

The Great Pacific garbage patch is now estimated to be twice the size of Texas, consisting of about 1.8 trillion pieces of plastic and weighing over 80,000 metric tonnes and (Lebreton et al., 2018). Given a world population of 8 billion, it is equivalent to 225 pieces of plastic for every person in the world. Its origin is not that of a few bad actors, dumping large amounts of plastic in the ocean, but rather hundreds of millions of individuals, each throwing small amount.  Although throwing plastic in the ocean is illegal in most countries, it is still hard to enforce. This in turn makes compliance dependent on individuals choosing not to do what might be most convenient, but rather what is most beneficial for others. That is, compliance is dependent on individuals participating in prosocial behaviour. In its absence, seemingly trivial individual actions can accumulate to large problems in aggregate. 

Fortunately, people regularly participate in voluntary prosocial behaviour. Blood donations, charitable giving and recycling are all examples of behaviours that incur a cost for the individual but benefit others. Even if these are common, they often fall short of the socially optimal level, resulting in many global problems, such as antimicrobial resistance, depletion of natural resources, and indeed, several oceanic garbage patches, for example, in the North Atlantic and the Indian Ocean.

What can be done to mitigate these problems? If prosocial behaviour is insufficiently incentivized, economics offers a simple solution: monetary incentive ought to improve the desired behaviour. Unfortunately, this seemingly ready solution does not always deliver. Sometimes incentives backfire and increasing monetary reward actually reduces the behaviour it was meant to increase, a phenomenon known as ‘motivational crowding-out’. The literature addressing when, how and under what circumstances crowding-out is likely to occur has yielded mixed results. The reason is likely that studies to date have faced a number of limitations, e.g., they have been of relatively small-scale or have only been able to estimate a few points on the relationship between reward and behaviour.

To overcome these shortcomings, we would ideally want an exogenous change in reward for participating in a common and naturally occurring prosocial behaviour, that spans many different reward levels. A good example is the recycling of beverage cans in Sweden; a prosocial behaviour that more than ninety-nine percent of the adult population engages in (Pantamera, 2022). When a can is bought, a small deposit, explicitly stated on the container, is paid at the time of purchase which is then subsequently refunded when the can is returned to the store and deposited in designated recycling machines. The researchers’ problem, of course, is the lack of variation in deposit (reward) that is needed to study the variation in recycling.

A rare opportunity presented itself when the Swedish central bank decided to remove the coin with the lowest denomination (SEK 0.5, equivalent to 0.05 USD at the time) from circulation. This exogenous decision forced manufactures of recyclable beverage cans, which carried a SEK 0.5 deposit, to gradually replace them with cans carrying an SEK 1 deposit. This was done gradually over three and a half years, but the rate of replacement differed greatly between stores, inducing the necessary variation for evaluating the relationship between reward and behaviour. In the end, the expected reward for the average recycling customer, recycling once a month, gradually increased from SEK 9 to 18 (approximately USD 1.25 to 2.5). Knowing this, we could utilize data on recycling from the relevant period to measure the effect. We were fortunate enough to already have good relations with one of the largest grocery retail chains in Sweden (Kooperativa Förbundet (KF) from a previous research project. They kindly provided us with detailed monthly recycling data from all their stores across Sweden, which resulted in a balanced panel consisting of 485 stores spanning the relevant 42-month. In the end, we had 20,370 store level observations based on aggregation of 27 million individual recycling decisions.

Using this information, we could see how recycling behaviour changed as expected reward increased, and the result was surprising. Initially, there was the expected increase in recycling when reward increased that economics would predict; but as reward kept increasing the effect on recycling started to become negative. This negative effect continued until expected reward became large enough such that the effect on recycling once again became positive. This means that the relationship between expected reward and recycling was “s-shaped” rather than strictly increasing (i.e., there was first a positive segment (the lower part of the s-shape), then a negative segment (middle part of the S-shape), and then finally a positive segment (the top part of the S-shape)). The effect size was substantial - moving from the first to the second turning point corresponded to a decrease in recycling of 89 million cans annually, across all Swedish stores. To put this in perspective, the reward would have to increase with more than 10 % on average to compensate for the crowding-out effect. This means that there was an opportunity to get more for less: it would be possible to achieve a higher level of recycling with a lower reward.

So, what have we learned from this study? A key insight is that care has to be taken when implementing monetary incentives targeting small individual actions, especially where only small rewards can be offered, e.g., deposits for plastic bags, return of unused antibiotics or, as in our case, recycling; very small differences in reward for seemingly small individual acts can have large accumulating consequences, and potentially lead to large downstream problems, as the Great Pacific garbage patch illustrates. We also believe that the main empirical finding, the s-shaped pattern relationship between reward and recycling, can help both reconcile the mixed literature on motivational crowding-out and underscore the need for careful testing before implementing large scale policies that include rewards for prosocial behaviours. 



Lebreton, L., Slat, B., Ferrari, F., Sainte-Rose, B., Aitken, J., Marthouse, R., Hajbane, S., Cunsolo, S., Schwarz, A., Levivier, A., Noble, K., Debeljak, P., Maral, H., Schoeneich-Argent, R., Brambini, R. and Reisser, J., 2018. Evidence that the Great Pacific Garbage Patch is rapidly accumulating plastic. Scientific Reports, 8(1).

Pantamera., 2022. Fakta & Statistik. [online] Available at: <https://pantamera.nu/sv/privatperson/fakta--statistik/> [Accessed 23 March 2022].

Conny Wollbrant

Professor of Economics, University of Stirling