Many studies have shown a relationship between antidumping duty and collusion. These studies, however, only focus on collusion in output (downstream) market, i.e. collusion between import competing firms and exporters, or among import competing firms. This dissertation explores how the antidumping duty on downstream goods can affect collusive behavior in an upstream market of exporters whom are sub jected to the duty.

 Bertrand duopoly model with infinite periods is developed to examine the effect of the antidumping duty on collusive behavior. Under a set of discount rate, whether is influenced by a tariff or the antidumping duty, the exporters will fully cooperate. The unaffected rate might be due to the linearity in input supply and output demand assumptions. Although the discount rate is not suffciently high enough to support the full cooperation, the collusive behavior is still feasible through self-enforcing agreement. With future period self-enforcing agreement, under the antidumping duty, the full cooperation in the initial period that is feasible under a set of the discount rate is called "the restricted full cooperation". The set under free trade that supports the full cooperation is smaller than the one supporting the restricted full cooperation. Therefore, the antidumping duty on downstream goods is pro-collusive in the upstream market. 

 The theoretical result is tested by using Thai shrimp industry data during 1996-2009; the industry has been sub jected to the U.S. antidumping duty since 2005. 2SLS is employed to estimate a system of Thai fresh shrimp supply, the U.S. demand for Thai frozen shrimp, and the mark up equations. Using comparative static in supply approach, with an interaction between fresh shrimp price and rainfall as a supply rotator, the empirical results confirm that the antidumping duty increases the degree of collusion among the exporters in Thai shrimp market at 1 % significant level.