The Determinants of Financial Analysts' Performance: Analyses using Quasi-Natural Experiments

Abstract

This thesis consists of three studies that utilize financial analyst career events as quasi-natural experiments to examine the factors that explain analyst forecasting performance. The purpose of this thesis is to minimize endogeneity problems that have hampered the financial analyst literature and at the same time add to the literature by showing that important life events can have a significant impact on analyst forecasting performance. First, I examine how employment change affects analyst herding behavior in their forecasts. My results show that analysts exhibit stronger herding behavior following an employment change. Specifically, they have a greater tendency to imitate other analysts’ earnings forecasts. Also, relative to their peers, they are slower in issuing forecasts and, as a result, issue revisions less frequently. This has a consequential negative effect on the market impact of their forecasts. I argue that the results are due to the need for newcomers to contend with the unfamiliarity of their new workplace environment and demonstrate that my results hold across several robustness tests, including a quasi-natural experiment using brokerage firm M&As that utilizes the estimation of an average treatment effect. This study raises a significant human resource question on how brokerage firms should support employees who have recently switched jobs. Second, I examine the impact that work specialization has on the performance of superior and inferior analysts. My results show that the forecast accuracy of superior analysts improves when their coverage is more concentrated within a few industries. However, there is no evidence of an equivalent improvement for inferior analysts. I argue that this is due to superior analysts being better able to utilize intra-industry relevant information when pricing stocks within the same sector, leading them to benefit more viii from specialization. My results are robust when I conduct quasi-natural experiments by utilizing brokerage firm M&As to capture changes to the work specialization of analysts who continue to work in the merged firms after the M&A events. The findings of this study have implications for how brokerage firms allocate coverage to analysts with different abilities. Third, I examine a channel that can explain analyst forecast pessimism. Specifically, I investigate the forecasting performance of analysts who have been rehired after experiencing a recent job loss following their brokerage firm closures and find that their forecasts will be more pessimistic relative to both their peers and actual earnings. Importantly, this leads to a decline in the accuracy of their forecasts at their new job. These results are theoretically supported by the career transitions literature, which shows that a job loss will affect the mental disposition of an employee and which I argue leads to analysts providing more pessimistic recommendations. This raises an important question as to how brokerage firms should support new employees who have recently experienced a job loss to avoid any negative impact it might have on their performance.