An empirical investigation into the association between the financial bath and income smoothing

A continuing concern of financial statement users is the comparability of financial information. Within the current framework of accounting principles, certain practices are possible which may reduce interperiod and intercompany comparability of financial statement data. One of the factors that may affect comparability is the postulated tendency of the managements of some firms to attempt to manage their reported earnings streams over time. Two of the most important of possible income management methods are the financial bath and income smoothing.

If an enterprise attempts to manage income by taking financial baths in certain periods, it may also attempt to manage income during the intervening periods by smoothing income. This assumption is the foundation for this project. By dichotomizing firms into those that may attempt income management and those that may not, it is asserted that enterprises taking financial baths also have smoother income streams during the intervening periods between bath years than enterprises not taking baths.

The general approach taken in this study is to select two samples of possible bath companies and a sample of non-bath companies (the control group) and to compare the dispersion around trend lines for l4 selected variables for each of the two bath company samples to the dispersion around the trend lines for the control group. One of the two bath company samples consists of companies disclosing bath items in their financial statements during the period l973-l976, while the other is comprised of companies that may be disguising bath items in aggregate financial statement categories during this period. The 14 selected variables consist of six income variables (such as income before extraordinary items and discontinued operations) and eight nonincome variables (such as dividends per share). Dispersion around the trend lines is measured by (l) coefficients of variation, and (2) the mean absolute percentage deviation of the actual changes in a variable from its trend.

The results indicate that there are differences between the properties of the bath company samples and the control group. For the bath company sample which disclosed bath items, the sets of dispersion-measurement values for the six income variables and the taxes variable are significantly greater (at the O.05 level) than those for the control group. The evidence does not support the assertion that there are significant differences for the remaining seven variables. For the bath company sample representing companies possibly disguising bath items, the sets of dispersion-measurement values for the six income variables and the taxes variable are significantly greater than those for the control group before elimination of the bath years. After elimination of the bath years, the evidence does not support the assertion that there are significant differences between the sets of values for this bath company sample and the sets of values for the control group for the l4 variables.

It is concluded that there are significant differences between the income streams for bath companies and the income streams for non-bath companies. In this study the assertion is tested that the

income streams for bath companies are smoother than the income streams for non-bath companies. The results indicate that the opposite conclusion must be drawn; the income streams for non-bath companies are smoother than those for bath companies. In addition, bath companies may add and discontinue operations more often than non-bath companies providing them with greater opportunity to use the financial bath. Greater fluctuations may be introduced into their earnings streams as a result.