My dissertation examines individual analysts' specialization in earnings forecasting skill and stock picking skill. I analyze analyst utility maximization process and predict that non-specialization is optimal for analysts when there is considerable economy of scope between the two skills' development; that specialization is optimal when the costs of skill development are high; and that the marginal benefit of each skill is positively correlated with the chance that analysts choose to develop that particular skill. Consistent with my hypotheses, I find empirical evidence that an analyst's choice of non-specialization is positively correlated with his brokerage size, brokerage reputation, his all-star status, the industry skill spillover effect, all of which capture the inverse of the costs associated with skill development. I show that the economy of scope in skill development, measured with earnings relevance and earnings timeliness, can explain an analyst's choice of skill specialization versus non-specialization, albeit in a non-linear manner. Using earnings persistence, earnings predictability, and earnings fixation to measure investors' demand for analysts' earnings forecasting skill, and firms' shareholder base to measure the demand for analysts' stock picking skill, I show that these firm characteristics explain analysts' choices between the two skill specializations. I also find analysts who specialize in earnings forecasting utilize the skill more intensively than analysts who do not by issuing more frequent and more innovative earnings forecast revisions. However, there is no significant relation between relative stock recommendation frequency and stock picking specialization probably due to the highly discrete nature of stock recommendations. My study not only empirically tests the theories of labor specialization but also improves our understanding of analyst utility maximization process and skill development.
展开▼
