Management Forecasts and Statistical Prediction Model Forecasts in Corporate Budgeting
比较了公司内部三种管理销售预测与回归、时间序列模型的准确性，发现统计模型优于人工判断，并揭示了定性因素和政治博弈对预算过程的影响。
In this paper, we examine, in a corporate environment, the accuracy of three management sales volume predictions relative to the accuracy of predictions from regression and time-series models. In addition, we explore the organizational uses of sales forecasts and budgets, the interplay of subgroups in trying to satisfy multiple objectives, and the sources of management forecasting error. Our analysis of sales volume forecasting errors (absent any knowledge of organizational influences or objectives of the decision makers) suggests that the firm can improve forecast accuracy by using statistical models instead of human judgment. This result confirms prior research which suggests that decision makers have difficulty in consistently applying their information processing strategies (Ashton [1984] and Libby [1981]). In interviews conducted to uncover the sources of prediction error, managers revealed that qualitative considerations and political maneuverings influenced the budgetary process. The company's accounting control system provided incentives for some managers to manipulate otherwise accurate sales volume predictions to satisfy conflicting budgetary objectives.