A Systems-Based Model and Auditors' Control Structure Judgments: Development and a Preliminary Test

Professional standards require that auditors develop an understanding of their clients' control structure forjudging how heavily the controls can be relied upon and for the determining of the type and extent of the subsequent tests to be undertaken. The auditor's judgment in this situation involves identifying and analyzing a complicated set of contingencies. This paper develops and tests a model that links the three elements of control structure: control environment, accounting system, and control procedures. The purpose of the model is to characterize the effects of several general classes of variables on control structure reliability. The model is based on concepts taken from a general systems theory called cybernetics. It has four components: environmental constraints (e.g., presence of formal ethical standards), systems boundary constraints (e.g., transaction approval), feedback regulators (e.g., monthly numerical reconciliation of invoices), and programmatic regulators (e.g., formalized procedures for investigating and correcting mismatched pairs of internal documents). The proposed model predicts that, consistent with current professional standards, all four classes of variables have direct effects on control reliability. The proposed model also predicts specific kinds of contingent and combined effects for pairs of components. Although current professional standards assume the existence of such effects, they do not specifically identify them. The model's predictions are compared to the judgments of 30 experienced auditors regarding the reliability of the control structures described in multiple case studies. The model is not viewed as a model of auditor judgment processes. However, to the extent that the model does an adequate job of characterizing the relations among the variables, its predictions should be consistent with experienced auditors' decisions. The results of the comparison provide evidence of strong support for the model's prediction that all four classes of variables are significantly related to control structure reliability. The evidence in this study also provides limited support for the model's predictions of specific contingent and combined effects of variables. Specifically, the model developed in this paper suggests that feedback regulators and programmatic regulators have an amplifying effect on the reliability of controls. This result is also shown in the auditor judgment data. Such a finding is significant because it establishes the model as a systematic means to predict nonlinear relationships in auditors' judgments. The model thus holds promise both as a way of describing control relationships and as a means to predict when auditors will demonstrate amplifying or compensatory effects in their judgments.