1994 Outstanding ABO Dissertation

Attributes of Dominant Control: Theoretical Model and Empirical Test

By Keith R. Duncan, Bond University

Abstract

The international trend in standard setting requires consolidation of all entities where the 'group' controls the operating and financing decisions. The question now facing the business community is "how do we identify the members of the extended 'group' for inclusion in the accounting entity." Identifying 'group' membership and the correct application of the three accounting treatments for investments--consolidation if controlled, equity method if significant influence, and lower of cost or market otherwise--is predicated on a clear delineation of the attributes of dominant control. Thus the research question for this thesis is:

What combination of attributes gives one entity "control" over the decision making of another entity?

Control is modeled as a continuous concept (Ci) which accountants arbitrarily trichotomise for reporting purposes. The upper cut-off, Oc' partitions dominant control situations from significant and insignificant influence relationships. Practitioners must assess the probability that the group exercises dominant control over another entity. For the ith control judgement this probability is modeled as a linear function of multiple control attributes (i.e. Xij, j=0, 1, 2, ..., n).

A review of the literature in accounting, law, finance, economic, and management suggests four key attributes determine the level of dominant control: (1) the total ownership level; (2) the level of direct versus indirect ownership; (3) the dispersion of nonowned equity; and (4) the level of representation on the Board of Directors.

This research employs a conjoint methodology empirically estimate the implicit weights subjects, consisting of businessmen, accountants, and financial analysts, place on these four control attributes. The quasi-experimental design presents subjects with preplanned scenarios that differ in the orthogonal combinations of ownership, directness of ownership, dispersion, and board membership attribute levels. Subjects judge the degree of control that they feel each situation depicts through conjoint analytic techniques these judgements are disaggregated to reveal the relative attribute weights for each individual, and for the respondent group as a whole.

The results indicate that ownership and board membership are perceived to be the most important attributes in dominant control relations. The implicit weights for indirect ownership links and dispersely held non-owned equity, suggests these attributes mitigate the level of dominant control achieved through ownership and board membership. The model is found to exhibit predictive ability within the estimation sample and for a set of holdout observations. The simple linear model performs at least as well as more complex ideal point, quadratic, and two-way interaction specifications. The estimation procedures are also robust to potential violations of their theoretical properties. Finally, the crosscultural instability of the estimated model parameters is found to be driven by the clustering of opinions within the two cultures in the sample (i.e. US and Australia).

The research provides a model, based on consensus professional judgement, that predicts whether one entity controls another. This model can be used by auditors, corporate accountants and regulators to assess control relationships. Further, researchers can use the model to test reporting problems that relate to the level of control. Finally the study represents an exposition on the conjoint methodology as a research tool for investigating multi-attribute decision making in accounting and auditing.