This paper considers a model of a rating agency with multiple clients, in which each client has a separate market that forms a belief about the quality of the client after the agency issues a rating. When the clients are rated separately (individual rating), the credibility of a good rating in an inflationary equilibrium of the signaling game is limited by the incentive of the agency to exaggerate the quality of the client. In centralized rating, the agency rates all clients together and shares the rating information among all markets. This allows the agency to coordinate the ratings and achieve a higher average level of credibility for its good ratings than in individual rating. In decentralized rating, the ratings are again shared among all markets, but each client is rated by a self-interested rater of the agency with no access to the quality information of other clients. When the underlying qualities of the clients are correlated, decentralized rating leads to a smaller degree of rating inflation and hence a greater level of credibility than in individual rating. Comparing centralized rating with decentralized rating, we find that centralized rating dominates decentralized rating for the agency when the underlying qualities are weakly correlated, but the reverse holds when the qualities are strongly correlated.