THE IMPACT OF TAX COMPLEXITY OF ANALYSTS' EFFECTIVE TAX RATE FORECASTS (CAPITAL MARKETS, INFORMATION PROCESSING)

Information complexity, which increases with the amount of attentional capacity or mental processing required to use information, is directly related to the cost of using the information. In this dissertation, I suggest that the complexity of information available to capital markets and the environment within which it is impounded varies across firms resulting in differences in the cost of incorporating that information. I employ a framework from the judgment/decision-making research to provide a link between complexity and analysts' forecast accuracy consistent with an information processing cost explanation for analysts' failure to use all information in forming earnings forecasts.; Prior research on how analysts form earnings forecasts presumes that the costs and benefits of incorporating information into forecasts do not vary across firms or by information type. Why analysts fail to use all available information has not been determined. This dissertation provides empirical evidence that analysts' choose to use information that is less costly to incorporate into their forecasts. Using a sample of analysts' forecasts from Value Line, implicitly holding the benefit associated with the use of information constant, I investigate the impact on analysts' accuracy of the variation in the costs of incorporating that information.; Using a measure of tax law complexity and analysts' forecasts of effective tax rates, I find that the magnitude of analysts' forecast errors is positively related to the firm complexity. I also find that analysts incorporate low complexity information (which is less costly to process) in their forecasts, but omit high complexity information (which is more costly to process). These results are consistent with an information processing costs explanation for the analysts' failure to use all available information and enhance our understanding of the impact of processing costs on information use.; The results reported, while consistent across various tests, are modest in economic terms. Due to data limitations, I include only firms with positive tax rates and earnings in the tests, limiting the generalizability of the results. In addition, results using earnings per share forecasts as the dependent measure are inconclusive, making it difficult to draw strong inferences about the implications of this study for prior research using earnings per share forecasts.