An Agency Interpretation of the Reliability Notion

1 The term reliability as an accounting concept has been in accounting theory before FASB. For example, in ASOBAT, similar notions, verifiability and freedom from bias, are emphasized as two of the four basic standards. 2 In an accounting setting, one can imagine a group of accountants are asked to evaluate the net income of a firm. Given any accrual accounting system, different accountants may arrive at different figures because they may have different bad debt or depreciation expenses estimates. In this paper we offer an agency interpretation of the reliability notion. In policy colloquialisms, the notion of reliability has been prevalent. In the FASB conceptual statement, reliability (together with relevance, the other primary quality of accounting information) is taunted as " central to accounting. " Further, reliability is decomposed into a representational faithfulness notion (correspondence or agreement between a measure or description and the phenomenon it purports to represent) and a verifiability notion (that several measurers are likely to obtain the same measure). 1 In accounting academy, Ijiri and Jaedicke [1966] provided the first theoretical formulation of reliability. Consider a group of n measurers who are asked to measure a given object. Under a given measurement system, difference measurers may produce difference numbers for the same object. 2 Let x i (i = 1, ..., n) be the quantity that ith measurer reports. Let x) be the mean of the sample, then a measure of objectivity may be defined as V = (1/n) G n i=1 (y i-x)) 2. The larger the V, the less objective the measurement system becomes. Intuitively, the measurement system is not very objective if different measurers may produce very disperse reports about the same object of interest. Let x * denote a theoretical " alleged value " of the object. A measure of reliability may be defined as R = (1/n)G n i=1 (y i-x*) 2. The larger the R, the less reliable the measurement system becomes. And it is elementary to see that R = V + B where B, for bias, is defined as B = (x)-x*) 2. Both objectivity and bias contribute to reliability. By definition, the reliability notion is intimately linked to x * , the " alleged value. " This is the key to understand reliability. Note the Ijiri/Jaedicke construction of the reliability notion corresponds nicely with the qualitative formulation offered by the FASB. Objectivity, measured by V, relates …