The Impact o f Dirty Surplus Flows on Performance Measurement
4.3 Measuring the impact of dirty surplus flows on abnormal performance
4.4.2 E V C absolute errors
Table 4.2 reports statistics and tests for cross-country variation in the EVC absolute errors. Absolute errors assess the inaccuracy in abnormal performance measurement. The null hypothesis that the distribution o f errors is centred on zero is not tested for the absolute values o f errors, as all values must be non-negative.
Contrary to the effects observed in the signed errors, which are largely due to goodwill-related items, all classes o f dirty surplus flows give rise to significant cross country differences in absolute errors. For both scaling measures, and for both horizon intervals o f three and eight years, the null hypothesis o f equality o f mean ranks in errors across the four countries is rejected for all classes o f dirty surplus flows. The tests o f equality o f mean ranks across pairs o f countries reveal that the level o f
40 For example, in the U.S. the FASB introduced, in force from June 2001, SFAS No. 141 Accounting f o r Business Combinations and SFAS 142 No. Accounting fo r G oodwill and Intangible A ssets, which prohibit the pooling-of-interests method o f accounting for business com binations and require a purchase accounting method where goodwill should be capitalised and depreciated (or subjected to periodic impairm ent tests).
rejections o f the null hypothesis, at the 5% significance level, is generally not sensitive to the scale measure used. For the three-year interval the only situation where the null hypothesis is not rejected simultaneously for the market value and EVC scale is for the ‘other dirty surplus flows’ category for the pair U.K./U.S. For the eight-year interval, this situation occurs for total dirty surplus flows for Germany/U.S., and for ‘other dirty surplus flows’ for France/U.K., Germany/U.S. and U.K/U.S.
Next, I analyse the level o f rejections o f the null hypothesis o f equality o f mean ranks across pairs o f countries for each o f the seven classes o f dirty surplus flows. For the three-year horizon, the hypothesis is rejected for all possible pairs o f countries in the case o f goodwill. For the EVC measure that disregards all dirty surplus flows, it is rejected for five pairs o f countries (France/U.K., France/U.S, Germany/U.K, Germany/U.S. and U.K./U.S). For the merger-related item, it is rejected for all pairs with the U.S. (France/U.S., Germany/U.S. and U.K./U.S.), reflecting the fact that merger accounting is mostly a U.S. accounting practice. For goodwill inclusive o f the merger-related item, the hypothesis is rejected for all pairs except France/U.S. For asset revaluations, it is rejected for all pairs except Germany/U.S., as asset revaluation are not permitted in these countries. For prior year adjustments, rejections occur for four pairs (France/Germany, France/U.K., Germany/U.K. and U.K/U.S.). Finally, for the ‘other dirty surplus flow s’ category the hypothesis is rejected for the five pairs (France/Germany, France/U.K., France/U.S., Germany/U.S. and U.K./U.S.) when the errors are market-scaled and four pairs (France/Germany, France/U.K., France/U.S. and Germany/U.S.) when the errors are EVC-scaled.
For the longer horizon o f eight-years, the null hypothesis o f equality o f mean ranks across pairs o f countries is rejected for all pairs for goodwill. For the category all dirty surplus flow, the null hypothesis is rejected for four pairs o f countries (France/U.K., France/U.S., Germany/U.K. and U.K./U .S.) in the case o f market-scaled errors and for five pairs (France/U.K., France/U.S., Germany/U.K., Germany/U.S. and U.K./U.S.) in the case o f EVC-scaled errors. The null hypothesis is rejected for all pairs in the case o f goodwill. For the merger-related item, it is rejected twice (France/U.S. and Germany/U.S.). For goodwill inclusive o f the merger-related item, it is rejected for all pairs except France/U.S. For asset revaluations, it is rejected for all pairs except Germany/U.S. For prior-year adjustments, it is rejected for the pairs France/U.K., Germany/U.K. and U.K./U.S. For ‘other dirty surplus flow s’, it is rejected twice (France/U.S. and U.K./U.S.) for errors scaled by market value, and three times for errors scaled by EVC (France/U.K., France/U.S. and Germany/U.S.).
In summary, regarding inaccuracy in EVC, the results show significant cross country variation for all classes of dirty surplus flows. Contrary to bias, inaccuracy in EVC is not only attributable to the omission o f goodwill-related items but to all types o f dirty surplus flows. Further, inaccuracy in EVC seems to occur for all accounting regimes and for different horizons. This is particularly important for business and managerial performance measurement based on residual income-type formulae. Disregarding dirty surplus flows will result in inaccurate calculations o f cumulative residual income over a multi-period interval. Consequently, using this measure to evaluate business performance, establish management remuneration schemes, determine value creation over a time-interval, or as a basis for any business decision may lead to incorrect assessments. Furthermore, using residual income-type measures o f performance that disregard dirty surplus flows in international comparisons may
result in misleading conclusions as results suggest cross-country variation particularly regarding inaccuracy.