Unusual patterns in reported segment earnings of US firms

Abstract

Purpose – The purpose of this paper is to extend prior findings on firms’ rounding up net income numbers to meet cognitive reference points and to examine whether segment-level earnings exhibit similar unusual patterns. Design/methodology/approach – This study is an archival research based on a sample of US public firms that report segment data between 1998 and 2011. The authors use Benford’s law to establish benchmarks for expected frequency of each number on the second digit of segment earnings and test whether the actual distributions deviate from expectations. Findings – The authors find more zeros and fewer nines than expected by chance in the second-from-the-left most digit for segment earnings numbers of US public firms, suggesting that segments round up earnings to meet cognitive reference points. The results complement the existing studies by showing that the rounding up of earnings not only exists at the firm level but also at the segment level, probably because subdivision managers have motivation to exceed certain reference points when reporting earnings. Research limitations/implications – As the authors cannot observe the contracts received by divisional managers, the authors rely on measures related to operating diversity to capture internal agency costs. Practical implications – The findings suggest internal and external auditors should pay close attention to segments that are suspected of earnings management, i.e. segments that report zeros on the second digit of revenues or earnings. Increased auditor attention is especially necessary for profitable segments operating in highly diversified multi-segment firms. Originality/value – The authors find that unusual patterns in segment reporting are more prominent in firms that operate in multiple and dissimilar segments, suggesting that higher internal agency conflict might lead to the rounding up of earnings.

Detecting earnings management using Benford’s Law: the case of Romanian listed companies

“…Geyer and Drechsler (2014) propose to check the correspondence with the BL for a balance sheet item -long-term debt -and find that US firms seem to manipulate numbers downwards. Garza-Gomez et al (2015) analyse the net income by business segments, presented by a sample of US firms, and find that the use of the numbers 0 and 9 in the second position moves away from the BL, in the case of profits; for losses, the two-digit frequencies approach the BL. Amiram et al (2015) calculate a score based on normal distributions of BL: FSD (Financial Statement Divergence score) and demonstrate that many of the indicators presented in the financial statements correspond to the BL, but the most likely to manipulations are the indicators of the income statement, while the least manipulated are the numbers presented in cash flow.…”