Does Superior Information Make Us More Discerning? What Uber Drivers Can Teach Us About Learning And Rationality
Communications Associate, World Bank
In 1957, Herbert A. Simon (Nobel Prize in economics 1978) introduced the concept of bounded rationality that recognizes that in decision making, human rationality is limited by the information we have, our own cognitive biases, our training and experience, and the finite amount of time we have to make a decision. Individuals and firms do the best they can with the information they have, and since they don’t have time to evaluate and rationally pick the optimal solution, they simplify their choices and go with one that is satisfactory rather than rationally optimal—this is calledstastificing.
Behavioral economics accounts for this by attempting to incorporate psychological insights. While most economists agree that there are some limits to the reasoning capabilities of individuals and firms, there has been much discussion about where and how to account for bounded rationality. On the spectrum between perfect rationality and the total absence of it, where are humans?
To explore this question, let’s take a look at cabdrivers and Uber drivers.
Economists have long argued how cabdrivers and other similar professions, like farmers or small business owners who regulate their own hours, decide how much to work each day. This question gets to the bottom of whether humans are fundamentally rational — in this case, whether they earn their incomes efficiently. The question has also taken on added importance as the gig economy, or shared economy, continues to grow and more people must decide how many tasks to perform each day.
On one side are behavioral economists who have found evidence that many taxi drivers work longer hours on days when business is slow and shorter hours when business is brisk— the opposite of what rationality would seem to prescribe. On the other side are more orthodox economists who affirm the traditional view of rationality argue that this irrational habit diminishes after analyzing more precise data that shows drivers generally work longer hours when business is good, allowing them to capitalize on fare returns (pun intended).
In a recent working paper, Michael Sheldon of the University of Chicago uses Uber’s proprietary data to analyze nine months of data in which independent drivers freely choose their hours and receive a fixed commission for every trip they complete. Unlike cabdrivers, whose hourly rate reflects how busy they are and does not change unexpectedly due to changes in fare prices, the hourly wage of Uber drivers reflects both busyness and rates, since Uber can increase prices when demand is high, also known as surge pricing. The paper’s main conclusion is that there was little evidence that drivers drove less when they could make more per hour than usual. Most Uber drivers tried to capitalize on surge pricing and worked longer when those prices were in effect.
However, this was not true for a large portion of new drivers who appeared to have an income goal in mind and stopped when they were near it, regardless of price changes. These drivers finished sooner when their hourly wage was high and to worked longer when their wage was low.
Sheldon did find, though, that this behavior decreased with experience, suggesting that income-targeting behavior, if present, was only temporary. “A substantial, although not most, fraction of partners do in fact come into the market with income targeting behavior,” but the behavior is then “rather quickly learned away in favor of more optimal decision making.” Greater experience teaches most drivers how to get the most out of their shifts and encourages them to abandon an income target, a goal not generally in their self-interest.
Essentially, Sheldon’s research suggests that rationality is developed, not innate — at least with regard to Uber driving.
Key to this idea of a learning curve is the provision of information. Cabdrivers, who must rely on rules of thumb that tell them, for example, that more people will call cabs when it’s raining or that there will be high demand around a sports complex at the end of a sporting event, and inexperienced drivers are more likely to fall back on an income target because they have less information about what they could be earning if they kept driving. In contrast, Uber drivers are informed when surge pricing comes into effect— sometimes for unexpected reasons like emergencies or traffic accidents.
So, it seems that humans can be found at different points on the spectrum between perfect rationality and the total absence of it, depending on how bounded their rationality is. We may start a new activity without enough information or expertise and fall back on heuristics, like an income target, to guide our behavior. This approach may not be optimal, but it satisfices our immediate goals. As we become more experienced and receive more feedback, our behavior evolves so that it hopefully converges towards rationality.