Legacy Institutions and Political Order in Weak States: Evidence from Chad
This dissertation investigates variation in the ability of non-state institutions to produce political order in weak states. In countries with weak central governments, non-state institutions, such as chieftaincies, are often seen performing many of the functions of a state: enforcing legal codes, collecting taxes, guaranteeing property rights, and ensuring security. However, while some chieftaincies demonstrate an impressive command over their followers, in other places, residents feel free to disobey their chief's edicts. I ask: Why do people in some places comply with their local chief, while in other places they do not?
Such chieftaincy institutions are often referred to as "traditional," "customary," or "informal," institutions. However these conceptual labels are inappropriate for the full range of institutions to which they are applied. Some of which have nothing to do with local traditions or customs, while some have written legal codes and official state recognition. Accordingly I consider them to be legacy institutions, based upon the historical legacy of their community.
I develop a theory of institutional time-dependent reputation and how it affects individuals' compliance decisions. I explain how centuries-old institutions can command greater compliance than newer institutions, because people grow up knowing the institution's reputation, believing they will be punished if they disobey its leader. In contrast, people are still formulating their beliefs about newer institutions, because they are unsure whether newer institutions are capable of following through with consequences.
I corroborate this theoretical argument with new evidence collected via immersive research in Chad. Using in-depth interviews with chiefs and a survey of 2,300 Chadian villagers across peripheral regions of Chad, I find that residents have higher expectations of compliance in areas where there are older institutions with established reputations. This finding that is robust to a variety of analytical approaches and statistical models.