Do forecasts produced by organizations reflect anchoring and adjustment?

Abstract

In attempting to improve forecasting, many facets of the forecasting process may be addressed including techniques, psychological factors, and organizational factors. This research examines whether a robust psychological bias (anchoring and adjustment) can he observed in a set of organizationally-produced forecasts. Rather than a simple consistent bias, biases were found to vary across organizations and items being forecast. Such bias patterns suggest that organizational factors may be important in determining the biases found in organizationally-produced forecasts, KEY WORDS Forecasting-organisational Behavioural decision-theory Anchoring AdjustmentIn attempting to improve forecasting, many facets of the forecasting process can be addressed. The statistical and other formal techniques used to produce the forecast can be improved. Allowance can be made for psychological biases of the individuals engaged in the forecasting process. The organizational context within which forecasts are produced can be altered. In this paper, we ask whether an extremely durable individual level forecasting bias (anchoring and adjustment) appears to characterize forecasts produced by organizations. An affirmative answer would suggest some very simple correction procedures might substantially improve forecast accuracy. In recent years, researchers in behavioural decision theory have demonstrated some extremely robust phenomena concerning how individtials make decisions under uncertainty and these empirical phenomena have been supported by interesting and succinct psychological theories (see for instance, Kahneman, Slovic and Tversky, 1982; Kahneman and Tversky, 1979). These researchers have done both positive and normative work on forecasting including biases in forecasts (see for instance, Tversky and Kahneman, 1974), confidence in forecasts (Fischhoff and MacGregor, 1982), and the process of developing forecasting models (Einhorn and Hogarth, 1982).In contrast to the individual focus of behavioural decision theory, this paper considers forecasts produced by organizations. Many, if not most forecasts, even if originated by individuals, are modified and approved by other members of the organization (Bromiley, 1986, Jenkins, 1982. In addition, the forecaster operates in a substantially more complex environment than the experimental subject and has access to numerous techniques and tools (e.g. statistical packages) that the experimental subject normally lacks. Organizational effects or the availability of formal tools might overwhelm behavioural decision theory phenomena in many interesting areas.

Predicting a firm's forecasting ability: the roles of organizational illusion of control and organizational attention

“…Exposure to limited alternatives means that the organization does not envisage alternate scenarios, resulting in sample bias. Insensitivity to outcome, elsewhere called adjustment and anchoring problems (Schwenk, 1984;Bromiley, 1987), refers to the imperfect readjustment of estimates despite receiving more accurate information (i.e., final estimates are biased toward initial estimates). All lead to inaccurate estimates and demonstrate a deficiency in organizational attention.…”

“…Where and how much relative to its competitors a firm focuses its attention impacts its forecasting ability (Schwenk, 1984;Bromiley, 1987;Das and Teng, 1999). Ocasio (1997: 204) stresses that 'the focusing of attention by organizational decision makers allows for enhanced accuracy, speed, and maintenance of information-processing activities, facilitating perception and action for those activities attended to.'…”

The Anchor and Adjustment Bandwidthmodel: Exploring Cognitive Biases in Simulated Forecast Series

“…This study applies four models defined by Bromiley [24], Harvey et al [25], Lawrence and O'Connor [26], and Amir and Ganzach [27] to identify and measure anchoring and adjustment (A&A) effects. These models provide a baseline for the Bandwidthmodels (BWM).…”

“…The first stated approach is defined by Bromiley [24]. Bromiley developed an approach to detect A&A on forecasts between one forecast (F), one actual, and one anchor value.…”

Managers and Analysts: An Examination of Mutual Influence

“…Managers may see themselves as better off easily exceeding a low analyst forecast rather than risking missing a higher forecast. While analysts will generally adapt to recent performance, they may reflect anchoring and adjustment by not adapting all the way to recent performance if that performance was well below their predictions (Bromiley, 1987). Issuing forecast guidance predicting performance below analysts' predictions then risks negative market reactions to the guidance, the subsequent adjustment of analysts' forecasts, and, potentially, subsequent actual low performance.…”