This is another (and probably the last) in our series of posts by PhD students on the job market – and one that is very close to home for those of us working in DC!
Financial incentives are an important policy tool for encouraging prosocial behavior. My job market paper  examines a simple element of incentive design – whether an incentive takes the form of a fee for bad behavior or a reward for good behavior – to assess how the framing of an incentive impacts the policy’s effectiveness. I determine whether small incentives and their design matter through the evaluation of two policies in the Washington Metropolitan Area aimed at reducing consumption of disposable grocery bags: a five-cent tax on disposable bag use and a five-cent bonus for reusable bag use.
Growing concern over the environmental impact of plastic bags has prompted several governments across the world to regulate the use of disposable bags; many countries in Europe, Asia, and Africa require grocery stores to charge a fee for each bag the store provides. In 2010, Washington, D.C. became the first city in the United States to pass legislation calling for grocery stores to tax customers five cents for the use of disposable bags. Two years later, Montgomery County, an area of Maryland bordering D.C., passed its own bag tax. In addition, individual stores could choose to offer their own incentives to discourage plastic bag use. In the Washington Metropolitan Area, several grocery stores offer a five-cent bonus for each reusable bag a customer uses instead of taking a new disposable bag. My job market paper asks the question: do these two incentives – a five-cent tax and a five-cent bonus –have the same effect on disposable bag use?
Standard economic theory suggests that financial incentives will be effective if the cost an individual associates with changing his behavior is smaller than the incentive provided for doing so. This implies that the form an incentive takes, a tax versus a bonus, should not impact the policy's effectiveness as long as the two incentives are the same amount (and that utility is linear in wealth over the relevant range ). If we assume that disposable and reusable bags are substitutes, both policies discussed above provide customers a five-cent incentive for using a reusable bag instead of a disposable bag. In contrast, evidence from the field of behavioral economics (Kahneman and Tversky, 1979 ) suggests that individuals perceive losses more strongly than gains of the same size. If grocery store customers are “loss-averse,” then a policy that charges customers for disposable bag use may be more effective than a policy that rewards customers for using reusable bags, even if the incentives are financially equivalent.
Several recent studies report the results of experiments that test the effectiveness of economic incentives with these behavioral mechanisms in mind (Field, 2009 ; Fryer et al., 2012 ). My paper contributes to this growing literature and provides new insights in a variety of dimensions. To my knowledge, this is the first paper to test for loss aversion in the context of environmental policy. Additionally, while the majority of papers that test for loss aversion in the field consider large incentives, this study provides evidence as to whether these behavioral findings hold with low-stakes incentives.
Does a very small financial penalty deter undesirable behavior? My empirical analysis begins with a review of previous studies on this question and provides evidence of the effectiveness of the tax policies in the Washington Metropolitan Area at reducing consumption of disposable bags. By observing customers as they exited the grocery store, I collected data on disposable and reusable bag use for over 16,000 customers at sixteen stores in the Washington Metropolitan Area including stores in Montgomery County (where there is a policy change), D.C. (which has a tax in both periods), and Arlington, Virginia (which has no tax) before and after the implementation of the Montgomery County tax. These data allow me to analyze the effect of the tax on demand using a difference-in-differences research design. While only 16 percent of customers in Montgomery County used a reusable bag prior to the tax, the reusable bag usage rate increased by 33 percentage points just two months after the tax was implemented – an overwhelming response.
Does a bonus for reusable bag use have the same impact on consumer behavior as a tax on disposable bag use? To address this question, I use cross-sectional variation in policies across stores to compare the effect of the bonus to the effect of the tax. The stores in my sample that offer bonuses are often located within walking distance of stores that do not and, therefore, the demographic characteristics of customers is similar across stores with different incentive policies. In stores that offer no incentive, 13 percent of customers use at least one reusable bag. This fraction is only slightly higher in stores that have a bonus (and not a tax). In contrast, 44 percent of customers in stores that have a tax (and not a bonus) use at least one reusable bag. These results suggest that, while the tax has a substantial impact on disposable bag use, a bonus of the same amount has almost no effect on behavior, evidence consistent with a model of loss aversion.
Loss aversion is not the only possible explanation for the observed difference in behavior across stores with different incentive policies. Alternatives include:
· Awareness. One reason the tax may have been more effective at changing customer behavior is that consumers may be more aware of the tax than the bonus. While survey data suggests that customers are less aware of the bonus than the tax, my analysis of this issue concludes that the differences in awareness cannot fully account for the difference in effectiveness of the two policies.
· Social Norms. I investigate whether the results are driven by customers responding to a shift in social norms associated with the tax. I surveyed customers on their attitudes about the use of disposable bags and pollution regulation before and after the implementation of the Montgomery County tax and found no significant change in response between the two periods.
· Tax Aversion. Recent evidence suggests that customers are more likely to avoid any charge that is framed as a tax (as opposed to a fee), i.e., they are “tax averse”. I conducted an online experiment in which participants were asked how they would respond to a hypothetical five-cent penalty for using a disposable bag, randomizing whether the penalty was framed as a government-imposed tax or as a fee instituted by the store. I found no difference between the two scenarios.
These results suggest that several current environmental policies could be improved by altering the framing of incentive. For example, Starbucks Coffee rewards customers who bring their own coffee mugs with a ten-cent discount. My results suggest that this policy might be more effective if Starbucks instead reduced the price of coffee by ten cents, but charged ten cents for using a paper cup. Similarly, the federal government awards a tax credit to customers who purchase environmentally-friendly Energy Star products. This policy might increase consumption of these products if they were taxed for purchasing energy-inefficient products.
I investigate the impact of a new “eco-sin” tax, a five-cent tax on disposable bags. I find that the tax policy reduced the overall demand for disposable bags by over half and prompted consumers to substitute with reusable alternatives. This is particularly notable given the relatively small size of the tax. The large effect of the tax is also striking in light of my finding that a five-cent bonus for reusable bag use had almost no effect on behavior, a result that is consistent with a model in which customers are loss-averse. My findings underline the importance of accounting for behavioral insights when designing a wide variety of incentives.
Tatiana Homonoff  is a PhD student in Economics at Princeton University.