Behavioral Economics incorporates psychological assumptions into the analyses of economic decision-making. The field of classical economics considers decision-making to be based on cold logic and hard facts – which human decision-making behavior does not always adhere to. Behavioral economics allows us to consider irrational behavior in decision-making and investigates the underlying reasons for these irrational choices.
Nudge theory, a core concept in behavioral economics, proposes subtle reinforcements and indirect suggestions as a way to influence the behavior and decision-making of groups or individuals. As defined by Thaler and Sunstein (2008), a nudge is an aspect of the choice architecture that alters people’s behavior in predictable ways without eliminating any options. Nudges are ideally easy and cheap to avoid, and are certainly not mandatory. An example of this can be found in the supermarket. Products can be arranged in a certain way to attract attention and encourage customers to buy certain products. A choice architect can choose to display healthy products at eye-level and place high-calorie foods in the lower shelves. In this case, the arrangement of the products in the supermarket is simply nudging the consumer to make the healthier choice without taking away the alternative, i.e., banning junk food.
The use of digital nudges has also risen over time – not particularly surprising, considering how much of an individual’s day is spent on the internet. This increase in time spent online allows us access to a lot more data about a customer’s preferences, purchases, lifestyle habits and so on, which in turn broadens the scope of applicability of nudge theory. Nudges are extremely apparent and are popularly used on retail websites to push for more sales, higher cart values, etc – some common nudges include notifications with social proof (e.g. 12 more people are viewing this product right now), decoy effect in product comparison (e.g. a graphic comparing three smartwatches in order to push sales for a specific smartwatch), opt-out shipping options and goal gradient during the checkout process.
Although nudges have been effective on the population level, the one-size-fits-all approach is characterized by weak generalization on the individual level. While nudge policies show a positive outcome on average, it is often impossible to determine how effective they are for specific individuals. A majority of public policy changes using behavioral economics uses a broad-brush approach. Behavioral economics studies often note their results as follows: 90% of people reacted in a certain way. However, this excludes 10% of the population – which can be problematic in policy interventions. For example, a broad intervention implemented in education can exclude and negatively impact populations of slow-learners or students with disabilities.
Recent developments in machine learning have the potential to alleviate this problem using models with high predictive power and lead to more personalized interventions. A generalized iterative learning model that can be applied to each individual separately and generate personalized predictions can be utilized to yield the desired outcome. Development nudges are effort-based and often nudge you to put in more effort, as compared to retail nudges which typically focus on effort reduction – this can likely explain the gap in the success of nudges in the development sector as compared to the retail sector. Personalized nudges can help us change that. For example, personalized policy interventions can be applied to a variety of niche areas such as rehabilitation, healthcare, nutrition, lifestyle habits, etc. Discussed below are some of the possible applications of precision nudging, particularly in development-oriented sectors.
Personalized nudges can be highly optimized when it comes to designing nudges for mobile-related behavior such as addictions centering around excessive cell phone usage, internet usage, online shopping, gaming, etc. Devices such as smartphones allow us to garner a large amount of data from particular consumers – this data can be used to tailor personalized nudges that prompt individuals to reduce problematic behaviors. For example, depending on the type of addiction, some individuals might respond better to periodic reminders that they’ve used the phone continuously for x minutes in a session whereas for other individuals a suggested break from phone usage during family dinners could induce certain behaviors during particular timeslots.
2. Nutrition and Lifestyle
Nutrition is a very personal concept – each individual has different nutritional requirements and would prefer tailored diets. Furthermore, an individual’s affinity to follow such a diet depends on a variety of factors – accessibility to certain kind of foods, work-life, activity level, place of residence, social conventions, health records, etc. Using these factors in a model, personalized nudges can be designed in order to prompt individuals to eat healthier and furthermore, be prompted to maintain healthier lifestyles – leading to reduced healthcare costs over the longer term.
A majority of hospital re-admissions post-surgery are due to non-adherence to discharge protocols. Models can be trained to predict which precision nudges would effectively serve a particular patient to reduce the probability of readmission. Hospitals often maintain extensive records of re-admissions – patient data can provide clues to a patient’s proclivity to follow different kinds of clinical recommendations (nudges). Nudges can also be used to serve as check-up reminders and / or medication reminders.
Nudges can be designed to prompt people to make greener and more sustainable choices. Environmental friendliness, for an individual, is dynamic, complex and highly contextual – it is contingent on a variety of factors such as cost, convenience, time, health, etc. Nudges need to be designed keeping this context in perspective. For example, people decide how to travel on the basis of a variety of factors such as time spent, cost of travel and convenience. A high toll, traffic jam, or parking fee can discourage people from taking a specific route or opt for using public transport Such a prompt on apps such as Google Maps alerting the traveler to these factors is likely to cause a behavioral change towards the greener side.
Such personalized nudges allow us to nudge individuals towards healthier choices and a healthier lifestyle with potentially high success rates. Although the abovementioned nudges are well-meaning and are designed with a common good in mind, the ethics of nudges has been called into question – there is a danger of manipulating people and presenting unfair nudges. Transparency, where users are made of aware of nudges used to influence them, is considered an important safeguard against manipulations and nudges that may not be in the user’s best interest – it also ensures that the autonomy of the users is safe-guarded at all times.
Hrnjic, E., & Tomczak, N. (2019). Machine learning and behavioral economics for personalized choice architecture. arXiv preprint arXiv:1907.02100.
Karlsen, R., & Andersen, A. (2019). Recommendations with a Nudge. Technologies, 7(2), 45.
Simpson-Young, B., Weeks, B., Sanders, M., & Minson, S. (2018). Behavioural Exchange: Big Data and Machine Learning. Presentation, Sydney. Thaler, R., & Sunstein, C. (2008). Nudge. New Haven, Conn.: Yale University Press.