Ko, Giovanni
(2012)
Competition, conflict and institutions: three essays in
applied microeconomic theory.
PhD thesis, London School of Economics and Political Science.
Abstract
This thesis consists of three papers on completion and conflict in three distinct but related
settings.
The first paper develops a model of tax compliance and enforcement where homogenous
agents receive signals about how tolerant the tax authority is of evasion, and where the latter
has imperfect means of detecting evasion. The main results show that increasing the quality of
the information that taxpayers have about the tax authority’s tolerance of evasion may increase
compliance. This is because if the signals are sufficiently informative, taxpayers are engaged
in Bertrand-like competition: if all taxpayers are evading a similar amount, each will have a
strong incentive to evade slightly below that amount in order to escape detection. This logic is
directly opposed to the culture of secrecy that prevails in many tax administrations.
The second paper, jointly written with Madhav Aney, deals with the question of how
specialists in violence like the military or the police can commit not to abuse their coercive
power. The answer that the paper provides is that competition between specialists in violence
creates incentives for them not to expropriate from civilians. The main theoretical results are
that these incentives become stronger as competition becomes more intense, both in terms of
the number of specialists in violence and in the evenness of their strengths. The hypothesis
that greater numbers of specialists in violence leads to less expropriation is tested using crosscountry regressions and found to be strongly consistent with the data, especially for the case
of developing countries.
The third paper analyses the equilibria of two-player imperfectly discriminating contests
of the power-form under incomplete information. This paper develops a method for solving
for the Bayesian Nash equilibria of such games by working backwards from the equilibrium
distributions of effort, rather than forwards from the distributions of the agents’ types. This
method is used to prove that there exist no distributions of type such that effort is an affine
function of the type. The method is used to construct an equilibrium where effort is loglogistically distributed, carrying out comparative statics. This equilibrium is shown to be
special in that it exhibits a formal equivalence to that in a contest with complete information.
Actions (login required)
 |
Record administration - authorised staff only |
Download statistics
Download statistics