Vets, pets and networks: What economics can teach us about our choices
Classic micro-economic theory seems to have fallen by the wayside as we all run after nudge behaviour and neuroscience, but it can still provide a useful lens when we look at how we are tricked into making choices that might be great for a business but not great for us.
In a nutshell, utility theory is an effort to explain why people choose one option over another, especially when the outcome is uncertain. It is the mathematical lovechild of 19th century utilitarians such as Jeremy Betham & John Stuart Mill, who came up with the rather revolutionary idea that ‘whatever makes the majority happy is the right thing to do’. Brilliant and simple, even though certain morally dubious, high profile lawyers have since used that idea to support things like torturing terrorists. Utility theory works quite well for simple, transactional decisions that involve easily measurable options. For example, paying $44 000 for a MBA degree makes sense if you think that you will earn an extra $20 000 per year over a career of another 25 years, with a 90% probability of completing the degree. So you would have a 0.9 chance of gaining $500 000 for a $44 000 investment. The expected utility of the degree is $450 000 — an almost 10x payoff.
But it’s not really that simple, is it? The expected salary increase is a risky assumption involving future employability, global economic health, expected lifespan, chances at academic success and many other factors that are entirely beyond one’s control. On top of that, by moving from 5 to 4 working days a week to make room for studies, what business or career opportunities might one be missing? Dealing with the potential losses from choosing one option over another is, of course, the opportunity cost. Beyond a certain point, it obviously becomes so speculative as to stop being useful. But the little escape hatch that allows us to quantify subjective value and deal with uncertainty is called bounded rationality. We don’t make fully rational decisions to optimise our payoff because we do not have perfect information and our decision making is massively influenced by contextual factors: physical (ever been hangry?), physiological (stressed vs relaxed, for example), societal bias (women aren’t good at engineering, for example) and others.
A classic 1976 Princeton experiment involving trainee priests showed how much contextual factors, in this case being time poor, influenced even the most fundamental moral behaviours. Further, people often use heuristics or ‘rules of thumb’ to make decisions, which often lead to suboptimal results. This leads us to behavioural economics and choice architecture, which is probably the most interesting part of microeconomics.
One of the interesting findings in behavioural economics is that we do not treat wins and losses in the same way, which is one of the main ideas of prospect theory. The theory was developed and popularised by Kahneman and Tversky, and the book itself can be spotted on many suburban bookshelves, although this tends to be for social signalling purposes, like many economics books. As an example finding, Kahneman showed that people have different behaviours towards risk when it comes to winning or losing. Given the same amount of gain or loss in terms of utility, the loss cuts us deeper. Winning $1000 feels good, but losing $1000 feels terrible. People therefore exhibit risk averse behaviour to avoid losing. This little human software bug is called loss aversion bias.
There are all sorts of other interesting findings in prospect theory. Another of these is how people go about selecting one option over another. There are two general phases: in the first editing phase, people subjectively decide which outcomes they consider equivalent. They then set this as a reference point and combine probabilities. This framing activity is then used to judge the outcome they expect or receive. In the second evaluation phase, people do some quick heuristics to see which option would lead to better outcomes/higher utility and then choose one option over another. They then use the non-rational reference point to decide whether not not they feel happy with the outcome of their decision. So-called narrow framing is used to explain how people are much more sensitive to gambles in isolation, but ignore many other adjacent risks, the classic example being the over-reaction of investors to the stock market’s changes, as opposed to changes in other investments.
We could therefore say that whereas classic utility theory put forward the idea of people as ultimately rational agents who choose the outcome with the highest value, the idea of prospect theory showed that this is not always (or even usually) the case. Prospect theory showed us that people do not always behave rationally. Sometimes, they make poor economic decisions. This is explained by our attitudes and biases towards certainty over uncertainty, loss aversion over gain seeking, the importance of relative positioning and a general discounting of small probabilities.
So let’s apply this to digital products and services, and specifically regarding the prioritisation of different features, the JBTD (job-to-be-done) framework. In a nutshell, different existing or potential features for products and services are mapped by the end users of the product: satisfaction against importance. Features that are important to the user and with which they are unsatisfied are prioritised for the next product improvement focus, as they are considered underserved. Served right and over-served are the two other segments that are either parked or possibly removed as features. This means that the expected utility value or importance is mapped against the fairly utilitarian value of satisfaction (happiness, perhaps?). Naturally, the interesting insights happen when one looks a bit deeper into the actual feature of the job: do you really want a 3mm masonry drill or do you want to look at a photo of your family on the wall? How much is that feeling of familial love worth to you? And how much would you pay for a frame and a drill and the time needed to put up the photo?
In another example from user experience design, the certainty bias is sometimes used by marketers to drive specific online behaviour. For example, customers may be more inclined to take action when offered a small, certain win over a large, uncertain win. This is often used to drive product reviews, the trialling of new products (which also helps overcome the default or status quo bias) or for loyalty programs that focus on retaining customers through loyalty discounts. The example above misses the target somewhat by only offering a possible large cash prize in exchange for a review. A more effective method would have been a guaranteed 5% discount voucher for the next purchase. Research shows that people are incentivised by the reward of the voucher, but fewer than 10% of customers actually redeem it.
Another culprit, and a fairly classic example, is the exploitative use of biases when creating demand for insurance products. Let’s use a Kiwi pet insurer as an example. Zoom, the dog on the far left, has developed a skin problem, which is fairly common — the cost to treat this is listed as $6000, which is unusually high. Archie the poor cat ate a piece of string and ended up costing his owner $4000. The type of incidents are both high probability occurrences to pet owners, but the associated costs are low probability. Our loss aversion bias will firstly be triggered to avoid the large losses and downplay the small, regular payments that we would make. We’d rather pay a bit now than deal with the large loss in the future. Further, these two incidents are far more likely to end up costing less than the website shows as ‘typical’.This example is actually doubly disingenuous: a high probability event has been linked in the customer’s mind to a small probability, high loss outcome(!). In fact, my field research with a representative sampling of one (at my local vet clinic) agreed that skin conditions are a high probability medical problem. However, she remarked that costs generally run into ‘a couple of hundreds of dollars, unless the owners decide to run all the tests and treatments, which they will only do if they have pet insurance’. This is a beautiful segue into the economic concept of externalities.
Externalities refer to the costs or benefits of a choice that affects a third party, who wasn’t necessarily involved in the choice. If we were to pull this into the standard price equilibrium concept, we could say equilibrium has been reached without the true costs being made explicit and factored into the transaction. In the pet insurance case above, we can see that pet owners with insurance will run all the tests and treatments without considering the cost beyond the initial co-payment, as this is the only cost they are incurring, while receiving outsized benefits. As the consumption costs are borne by all the other pet insurance owners in the form of increased monthly premiums, this is referred to as a negative consumption externality. Another well known example of this is the cost of environmental damage through pollution caused by the production of products. Any machine that runs on fossil fuels causes pollution, which is not factored into the cost of the production and sales price of the machine, unless through carbon taxes or similar penalties. These are referred to as negative production externalities.
But externalities are not always negative: renovating my house in Wellington will likely add some value to my neighbours’ properties as well, which is an example of a positive consumption externality. Or having my dog neutered and vaccinated will decrease vet costs through disease prevention, decrease animal suffering and create fewer unwanted puppies. This would increase the scarcity of puppies and increase the price and possibly profit margin for dog breeders, boosting consumer consumption and GDP. This is an example of a positive production externality.
These positive externalities are also common in digital networks, including all social media networks. Platforms such as Facebook, Instagram or Reddit exist as network phenomenons that are essentially positive externalities at scale. The classic dilemma for these platforms is the feedback loop problem: nobody will join a social platform if there aren’t many other members, who in turn don’t join because there aren’t enough other members. The most common remedy to this is the growth hack approach: finding a solution that will boost members rapidly enough to reach the tipping point of mutual benefit to the newcomer and network. Mark Zuckerberg understood this problem in the early days of Facebook, and solved it by hacking into the Harvard student database and creating personal profiles that students simply had to claim. It solved the problem, albeit in an unethical way, by showing these joiners that all their friends were already on the site (even though it was only an unclaimed name, very often). It was enough of a growth hack to kickstart the positive externalities inherent in the network effect.
In summary, a combination of our choices and random events (‘states of the world’) create moments when we have to make decisions of choosing one outcome over another, often by using a product or service. We do some quick mental math via heuristics to figure out which will be less risky and more beneficial to us now and in the future (‘expected utility’ or ‘payoffs’). We are not perfectly rational, as we are filled with biases and often dodgy software. We change our minds all the time, sometimes for transient reasons (contextual factors). But we are predictably irrational to a certain extent. We dislike losing more than we like winning, we worry less about effects that are removed from our immediate view and we use pretty subjective benchmarks about outcomes to decide whether or not we should be happy with our choices.
We might not be homo economicus, but we are not homo temere, either.
