Path Dependence

Path dependence, a highly popular concept in social science, simply refers to a dynamic pattern or continuity that evolves as a result of its own past. As a historically sensitive approach, path dependence emphasizes the role of the timing and sequence of events in the social and political world. Small, accidental, or random occurrences happening at a certain time are expected to have long-lasting, self-reproducing patterns or paths. In other words, when things happen in a chain of events affects how they take place. Below, the importance and also some of the limitations of this concept are discussed.

Although path dependence simply asserts that particular choices and events that occur in the past mold the unfolding sequence of events (i.e., history matters), any continuity or sequence of events may not constitute a path-dependent process. Path emergence, maintenance, and breaking in a path-dependent process should have specific features. Regarding path emergence, contingent, accidental, brief occurrences or nonoccurrences launch event chains. In other words, initial conditions are treated as stochastic. Otherwise, path dependence would be nonfalsifiable because one can easily link any event to an antecedent or a temporarily distant cause. As a result, any outcome might be marked as path dependent. To avoid such a problem of endless flow of causally connected events (i.e., the trap of infinite regress), the historical event that launches the path is expected to have properties of contingency.

Once historical contingencies or junctures set into motion certain patterns or sequences, they lock themselves in, leading to inertia. In other words, paths become inflexible or rigid over time, making it difficult to shift to another path or return to the initial conditions. The processes or mechanisms that reproduce a path might be rather different from the processes producing the path. Although stochastic factors start a path, certain causal mechanisms reproduce it. A highly emphasized mechanism of path reproduction is increasing returns, defined as self-reinforcing positive feedback processes. The idea of increasing returns suggests that the growing benefits that a certain path cultivates with its continued adoption create further incentives for path maintenance, leading to dormancy over time. Learning and coordination effects, adaptive expectations, and large setup or fixed costs constitute some sources of positive feedback processes. Thus, a path may reproduce itself even in the absence of the dynamics and factors responsible for its creation.

With respect to change, path dependence expects it to be incremental and evolutionary (i.e., path following or change within a path). Substantial, pathbreaking changes (i.e., change to a new path) rarely happen after long periods of continuity. Such major changes require critical junctures or periods (also known as branching [Page 1831]points) in which some externally driven contingent events, such as wars, economic crises, dramatic technological developments, natural disasters, or epidemics, make the extant path less attractive to follow. Thus, critical junctures unlock the existing path or equilibrium and generate a new one, leading to a new stasis (also referred to as punctuated equilibrium).

Pólya urn process, a simplified mathematical illustration, delineates the logic of path dependence quite well. Imagine an urn containing two colored balls, one yellow and one red. Let us draw one ball and then return it to the urn with an additional ball of the same color. If we repeat this process until we fill the urn completely, which balls will be the most frequent ones in the urn? Obviously, to a great extent this depends on the outcome of the initial draws (i.e., initial conditions). If we draw the red ball in the first round, then the likelihood of drawing a red ball increases in the following draw. Over time, the distribution would settle down to a majority of red balls. In other words, the ratio will eventually reach an equilibrium (i.e., the red one). This process indicates that sequence matters. In the beginning, we cannot say with certainty which color will lock itself in. We, however, know that the results of early draws in the sequence will have a substantial impact on which of the balls will be the dominant one at the end. Thus, early developments or events not only involve a substantial degree of randomness but also produce a large, determinative impact on subsequent developments. In other words, the impact of random occurrences early in a sequence of events does not cancel out. Instead, they exert a long-lasting influence on the evolution of the sequence of events (also known as nonergodicity).

...