Analyst forecast accuracy: Do ability and portfolio complexity matter?

摘要

This thesis investigates the association of security analysts' ability and portfolio complexity with their forecasting accuracy. Ability is defined as the amount of effort required to produce a forecast of a given level of precision. An analyst is said to be of higher ability if he requires less effort to produce a forecast at a given level of precision than other analysts. Portfolio complexity is defined as the amount of effort required to produce forecasts of a given average level of precision for a set of firms. A portfolio is said to be more complex when, holding ability constant, more effort is required to produce forecasts at a given average level of precision relative to other portfolios. This study uses proxies for ability that include employer size and years of experience and proxies for portfolio complexity that include the number of firms and industries followed. The use of these proxies is based on the assumptions that high ability analysts have more experience and work for larger firms, and that it requires more effort to forecast earnings for a larger number of firms and industries.;The hypotheses predict that high ability analysts and analysts who follow less complex portfolios will supply more accurate forecasts than other analysts. The hypotheses are tested using the I/B/E/S data set. I measure forecasting performance by comparing analysts' absolute forecast errors to the forecast errors of two benchmarks; the average analyst's absolute forecast error and the expected absolute random walk forecast error. I find that analysts' absolute forecast errors improve with experience. I also find that analysts who are employed by large brokers have absolute forecast errors that are smaller than other analysts. Finally, I find that expected absolute forecast errors increase with the number of firms and industries followed.;This thesis also investigates analysts' performance in the banking industry, with a focus on the role of specialization. When specialization is defined as the number of banks followed, with specialists having a higher proportion of banks in their portfolio, I find that specialists perform better than other analysts.