Small Firms' Forecast Rationality

Abstract

While there is a substantial literature investigating factors that cause or encourage firm growth, particularly for relatively smaller and younger firms, there is a dearth of research investigating the entrepreneurs' expectations and forecasts of firm growth and their degree of accuracy. This is despite the potential importance of forecasting. From an intra-firm perspective, forecasts and expectations have a vital influence upon many commercial decisions and subsequent firm growth, profitability and survival. Potential consequences of inaccurate forecasts include higher inventory costs, poor customer services, and inefficient utilization of production resources. There is also a need to understand forecasting behaviors and outcomes because of implications beyond the firm. The rationality of forecasts, efficient use of available information and trends by insiders influences planned and actual output, investment and the entry and exit of goods and services within industries. Consequently there are important economic theory and macroeconomic policy outcomes to be derived from understanding the forecasting process. This paper investigates the rationality of income forecasts made by small firm managers utilizing a large (3,500 firm) sample surveyed over a three-year period. Using statistical procedures common in the forecasting research literature, the following aspects of the accuracy and rationality of forecasts by small firm managers are examined: a) the degree of bias in forecasts; b) the degree of accuracy in forecasts; c) the degree of under or over reaction in forecasted growth; d) the degree of temporal and information set dependence in forecasts; and, e) comparison of individuals' forecast accuracy to the predictive accuracy of mechanical models. The major findings include: (i) inconclusive evidence concerning whether forecasts by managers exhibited optimism; (ii) an apparent propensity for managers to make forecasts that were more volatile than actual growth exhibited; (iii) the indication that, of the naïve time series models considered, random walk was the most accurate; and, (iv) the indication that compared to time series managers had superior forecasting ability. In summary, while systematic over-estimation of income was not found, managers did tend to make forecasts that were generally too extreme and tended to over-extrapolate previous growth.