The forthcoming article “Learning traps” by Hassan Sayed is summarized by the author below.
When do politicians experiment with novel policies? A mayor, for instance, may want to build a subway, but her constituents may not know whether this would be better than adding lanes to an existing highway. Or, a governor may want to introduce a new wealth tax, but voters may not know what effects this will have on their local business environment. The only way to learn about new policies is to see their effects play out in real life. Thus, even if a politician wants to experiment, she has to consider what constituents will learn.
This paper uses a repeated model of political agency to answer this question in the setting of long-lived leaders — like dictators or term-unlimited politicians. It shows that, when leaders are truly unsure about what benefits constituents, they pursue learning traps: half-baked, middle-ground policies that make it difficult to infer what policies are good or bad.
What makes learning traps uninformative? If our mayor fully expands the subway, voters will surely learn whether it’s effective. However, suppose she only partially funded the subway, and the results were mediocre. Proponents of the subway will say it isn’t working because it’s underfunded, while opponents will say it’s dysfunctional and overfunded. By mixing elements of different policies, learning traps make it hard to figure out what actually works.
What constituents learn today affects what long-lived politicians can do in the future. If a politician experiments with a policy she likes and it goes well, she can implement it for the rest of her tenure. But if it backfires, her hands are forever tied. Learning traps avoid this because constituents never have enough information to constrain her. By contrast, a short-lived leader — like a governor who only serves two terms — may have different incentives to experiment if she is worried about reelection, or do whatever she wants in her final term in office. A key contribution of the paper is thus to contrast the behavior of these “short-lived” and “long-lived” leaders.
What real-world phenomena can the model help us understand? Much of the paper applies learning traps to one of the largest reforms in European history: the end of serfdom in Imperial Russia. Prior to the 1860s, the Russian economy ran on a system of coerced agricultural labor known as serfdom. However, a series of geopolitical events in the late 1850s raised questions about whether alternative economic systems might serve Russia better. While Tsar Alexander II’s government seemingly preferred “liberal,” Western European economic institutions, they didn’t know if they would work in Russia. The ensuing reforms mixed communal and private property, stifled labor flows, and impeded Russia’s economic progress. The model lets us read this middle ground as a learning trap: even though the Tsar’s government preferred liberal reform, they pursued a middle ground to avoid worries of losing power in the future. The paper also helps explain regional differences in policy experimentation, why countries wage limited wars, and why politicians, despite their greatest ambitions, pursue tepid reform.
About the author: Hassan Sayed is a postdoctoral scholar at the Market Shaping Accelerator, a joint initiative between the Center for Global Development, Dartmouth College, and the University of Chicago. Their research “Learning traps” is now available in Early View and will appear in a forthcoming issue of the American Journal of Political Science.