International harmony measures of accounting policy: comparative statistical properties

Van der Tas's (1988) I index and the between-country C index introduced by Archer et al. (1995) are competing measures of international harmony. We present comparative statistical properties of these indices, via a simulation study covering three accounting methods in 10 countries, with uniform, bimodal and unimodal distributions of companies across accounting methods. The indices are also adjusted for non-disclosures using techniques developed by Archer and McLeay (1995) and Archer et al. (1995). The I index and the between-country C index are mathematically equivalent in the two-country case even in the presence of non-disclosures. As more countries are compared, the two indices diverge. The means and standard deviations of the I index, with a correction proposed by Archer and McLeay (1995), decrease and there is little skewness or kurtosis. In contrast, as more countries are compared, the between-country C index exhibits more stability in means, lower standard deviations, higher skewness and kurtosis. The between-country C index may be superior to the corrected I index because (i) between-country C index means approximate their ‘expected values’ (where all observations equal expected values) more closely than do corrected I index means: and (ii) between-country C index means are more stable than corrected I index means where the data come from stable distributions.