Bishnupriya Gupta, University of St. Andrews

This paper analyzes the International Tea Agreements of 1930 and 1933, whereby producers in India, Ceylon and Java/Sumatra agreed to reduce output and exports. The agreement of 1930 was not legally binding and the agreement of 1933 was given legal backing by the middle of 1933 in Ceylon and in Java/Sumatra but not in India, implying that repeated game mechanisms are likely to have a played a role in the enforcement of these agreements. We use output data for the period 1929-33 from sample of 349 British owned firms, in order to test the extent of compliance by firms to the international agreements. These firms are mainly located in Eastern India and Ceylon (with a smaller number in south India and Java/Sumatra) and account for a major proportion of tea output in these two important tea growing regions. Our main finding is that there was a reduction in output in both these years, indicating that collusion was successful. However, we also find regional variation. In 1930, the year in which the agreement had no legal backing in any of the countries, there was a significant difference in the extent of compliance across countries, with the firms located in Ceylon not reducing output to the extent of those located in India. In the year 1933, legislation to enforce quotas was introduced early in Ceylon and the Dutch Indies, but not in India. For this year we find no significant difference in compliance between the firms in Eastern India and Ceylon. This suggests that non-cooperative enforcement of collusion (via repeated game mechanisms or social sanctions) was more effective in Eastern India than Ceylon, so that legislative backing was necessary for a truly international cartel. We attribute this difference in ability to sustain collusion to differences in the organizational structure of the industry in the two countries Ð the degree of monopoly was higher in India, since firms were controlled by a smaller number of managing agents than in Ceylon. The paper also uses archival material on the negotiations leading to the agreements, as well as on the failure to collude in 1931 and 1932. This material indicates that the failure to collude in these years was mainly due to a bargaining conflict between established producers and the newer plantations of Java/Sumatra and should not be interpreted as the punishment phase of repeated game equilibrium.