We study the effects of trans-Atlantic passenger shipping cartels on tourist/business and migrant traffic. Collusion had a smaller effect on first and second class service relative to third class service. Its effects were proportionately stronger eastbound, but less important in absolute numbers given smaller eastbound traffic. Collusion-driven consumer substitution across classes was small but non-negligible. Thus, collusion affected migrant traffic far more than tourist/business traffic. We also confirm that collusion led to higher fares across all cabin classes, especially for steerage. We construct and calibrate an analytical model and show that the pattern of observed prices and volumes are consistent with a profit-maximizing cartel, thus buttressing the hypothesis that the collusion effects were causal. Finally, we document that collusion led to positive selection of migrants, as measured by literacy rates and financial resources.