![Sample our Area Studies journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5922/TOC-Subject-Sample-2021-Area-Studies.jpg"})
ABSTRACT
This study examines managers’ use of discretion in determining goodwill impairment losses and the differential value relevance of impairment information depending on the type of management discretion under IFRS in Korea. We distinguish management discretions into accelerated, timely, and delayed recognition of goodwill impairment. We find that impairment decision is driven by managerial incentives, such as big bath behaviour, income smoothing, and avoidance of loss reporting. Moreover, the impairment loss is positively (negatively) associated with contemporaneous share price and future expected cash flows, when managers recognize impairments in an accelerated (timely) manner. The positive association of accelerated impairment coincides with the negative value relevance of pre-impairment goodwill balance. We also observe that the accelerated impairers exhibit superior market performance in the 2-years subsequent to the reporting of goodwill-related information. Overall, the evidence documented in this study is in line with the positive valuation implication of accelerated discretionary impairment, consistent with a signalling perspective.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. The IASB conducted a post-implementation review of IFRS 3, to reconsider whether the impairment-only approach is functioning as anticipated, to see if it has improved financial reporting, and to identify areas of focus which warrant further investigation (IASB 2015). The IASB concludes that a review of academic research provides evidence which generally supports the current requirements particularly in relation to the usefulness of reported goodwill and its impairment, albeit some other studies show the impact of managerial incentives on impairment recognition.
2. Lee (Citation2011) finds that goodwill impairment charges in the post-SFAS 142 period are positively related to cash flows in the subsequent year. This, however, is not clearly elucidated.
3. Prior value relevance studies suggest that the book value of equity is a value-relevant factor that proxies for expected future normal earnings (Ohlson Citation1995). Similarly, it is argued that earnings reflect information about expected future cash flows (Kothari and Zimmerman Citation1995; Ohlson Citation1995). Consequently, the book value of equity and earnings are positively related to stock price.
4. We assume that investors rely on financial information at the firm level to estimate the fair value of goodwill and anticipated goodwill-impairment loss. The firm is treated as a single reporting unit, as though it has been purchased in a business combination. The market value of equity is assumed to proxy for the fair value of the reporting unit. The fair value of net assets is assumed to be equal to a firm’s book value. Then, the goodwill-impairment test can be simplified to the difference between the market value and the book value of stockholders’ equity. This line of logic leads to the market-based impairment indicator, as proposed by Ramanna and Watts (Citation2012), which is used to discern the normal from the discretionary decision group in this study.
5. Dichotomous classification based on two consecutive years’ BTM>1 could be an oversimplification. We employ alternative hurdles, such as below median or first quartile of BTM, to define a market-based indication of non-impairment. Non-impairment firm-years with BTM<median or first quartile are classified as a normal non-impairer group. In sensitivity analyses employing these alternative criteria, we find that the test results remain qualitatively similar.
6. Alternatively, we also employ accounting-based impairment indicators such as negative change in income from operations, consecutive deceases in ROA, etc. as a robust check. Li and Sloan (Citation2017) propose a higher percentage of goodwill balance coupled with a lower ROA as an impairment indicator.
7. We scaled reported impairment losses by other deflators such as total assets, and the results remained unchanged.
8. The definition of variables used throughout this paper is summarized in Appendix A.
9. This type of earnings management may benefit shareholders, if managed accounting information conveys signal about a firm’s future performance. Managers may use their accounting discretion to signal private information about better future performance (e.g. Healy and Wahlen Citation1999; AbuGhazaleh, Al‐Hares, and Roberts Citation2011).
10. There seems to be little additional penalty for lowering earnings that are already negative. Also, write-downs taken in the current period are more likely to induce improved future performance as the burdens of deficits are mitigated, and/or accruals are reversed.
11. We set the interval between 0 and 0.02 as the level of earnings likely associated with the motives to manage earnings upward to avoid loss reporting. We also experimented with varying intervals such as [0, 0.01], [0, 0.03] and obtained similar results.
12. In addition, we include other control variables used in prior researches. LARGEOWN, MGTOWN, FOROWN are the percentage of equity shares owned by the largest shareholders and related parties, executive managers, and foreign shareholders, respectively. BIG 4 equals 1 if the firm is audited by a Big 4 auditing firm and BODINDEP measures percentage of independent non-executive directors on the board. We expect that managers are less likely to engage in opportunistic earnings management with stronger corporate governance mechanisms as measured by these variables. SIZE is the logarithm of beginning total assets and GWTA is the ratio of pre-impairment goodwill to total assets. BTM measures the ratio of book value to market value of equity. Prior researchers find that companies which reported impairment losses in consecutive prior years are more likely to report an impairment loss in a given year (e.g. Elliott and Hanna Citation1996; Francis, Hanna, and Vincent Citation1996; Beatty and Weber Citation2006). We include variable YEARIMP, to measure the number of consecutive years with goodwill impairment losses before the current year to control for this effect.
13. To assess the statistical significance of differential value relevance of goodwill impairment, we introduce an indicator variable DISC, which equals 1 if the impairment decision is discretionary, i.e. accelerated or delayed, and 0 otherwise. We estimate OLS regression models which include interaction terms between IMP (GW) and DISC, whereby the differential slope coefficient on IMP*DISC (GW*DISC) captures the differential value relevance of impairment (goodwill) between normal and accelerated impairers (normal and delayed impairers). Provided that the decision is motivated by private information signalling about optimistic future prospects, the differential slope coefficient on IMP*DISC is expected to have a positive sign. This is because larger impairment losses recognized this period more likely bring about higher future performance, owing to big bath accounting, income smoothing, restructuring, economic turnaround, etc.
14. In additional analyses, we also employ various alternative procedures to mitigate the effects of influential observations. The results remain unaltered.
15. We also analyse the proportion of goodwill impairment in pre-impairment goodwill for firm-observations reporting goodwill impairment losses. From 2011 to 2016, the total amount of reported goodwill is 8,338 million KRW of which 2,274 million KRW of goodwill impairment losses (27%) are written-off. For the KOSPI sample, 1,191 million KRW (19%) is written-off from 6,322 million KRW of reported goodwill. For the KOSDAQ sample, 1,083 million KRW (54%) is impaired from 2,016 million KRW of goodwill, indicating higher percentage of impairment in KOSDAQ. Oil refining, rubber, plastic, chemistry industries show the highest proportion of goodwill impairment (44%), while the food & beverage industry shows the lowest (8%). In terms of market, the fiber, timber and pulp industry represents the highest (53%) in KOSPI and wholesale & retail industry represents the highest (88%) in KOSDAQ.
16. We also compared normal impairers and discretionary non-impairers to confirm that the latter has a higher frequency of LOSSAVOID and better economic performance compared to the former. Moreover, comparing the discretionary impairers and normal non-impairers reveals that the former is characterized by a higher frequency of BATH, CEOTURN, and lower performance relative to the latter.
17.. We also conduct an OLS regression, which obtains qualitatively similar results. For our OLS model, we employ a stepwise regression to reduce statistically redundant variables.
18. This explanation is supported by a relatively low mean value of 6.64% for the proportion of variable income to total income. Nonetheless, we should note that the lack of significant association might be driven by noise in relation to incomplete data. Public disclosure of CEO cash bonus information was mandated from 2013 in response to public demands in Korea.
19. When we estimate regression models for discretionary groups, normal observations serve as reference groups. Discretionary impairers are combined with normal non-impairers, while discretionary non-impairers are combined with normal impairers to highlight the incentives underlying each of the discretionary decisions to recognize or not to recognize goodwill impairments.
20. This result provides a clue to such questions as to why certain firms engage in downward earnings management when they actually perform relatively well, and whether such conservative accounting is interpreted by market participants as an optimistic signal about better future prospects. With regard to the second question, a valuation test is essential to facilitate meaningful interpretations.
21. Unlike other tests, we define the dependent variable in the probit model estimated for non-impairers column equal to 1 if an observation belongs to a discretionary non-impairment group, and 0 if it belongs to normal impairment group. Therefore, predicted signs of explanatory variables in the third column are opposite to those in other tests.
22. This type of accounting discretion is likely to exacerbate information asymmetry between the firm and external stakeholders. The concerns raised from practitioners and academics about impairment-only approach mainly relates to this situation.
23. Based on a sample of 2,123 acquisitions by US companies in the years 2002 to 2006, Lys, Vincent, and Yehuda (Citation2012) identify 1,251 acquisitions that resulted in economic profits and 871 acquisitions that resulted in economic losses for the acquirers. They find that in the acquisitions subsample with economic losses the disclosed goodwill is negatively correlated with the losses. The authors suggest that the reason for the negative correlation is that the recorded goodwill includes overpayment by the acquirers which should have been written off at the date of the business combination.
24. In regression analyses to assess the value relevance of goodwill impairment, we use absolute value of impairment losses to facilitate interpretation, following prior literature. The negative (positive) sign of the absolute value term indicates that impairment loss reflects reduction (increase) in expected future economic benefits incorporated in current share price.
25. Realized 1 year and 2 years ahead cash flows are also considered as alternative metrics. The dependent variable should only contain the cash flow that is expected in year t, thus, the realized cash flows are subject to measurement error since they contain both expected and unexpected components. The errors can be mitigated by including future returns ( in our model, which are the proxy for the measurement error (Jarva Citation2009).
26. Further, to test the differential predictive ability of goodwill impairment information between normal and discretionary impairers and non-impairers, we extend the baseline model by including interaction terms with a DISC dummy variable that equals 1 if the decision is discretionary, and 0 otherwise. The results suggest that discretionary impairer group exhibits significantly more positive association between impairment loss and expected future cash flows.
27. The first-stage choice model distinguishes impairers from non-impairers. Heckman’s approach is applied both for value relevance and predictive value tests.
28. Earnings management is defined as white when it is taking advantage of the flexibility in the choice of accounting treatment to signal the manager’s private information on future cash flows. Black earnings management refers to the practice of using tricks to misrepresent or reduce transparency of the financial reports. Grey earnings management is choosing an accounting treatment that is either opportunistic (maximising the utility of management only) or economically efficient (Ronen and Yaari Citation2010, 25).
29. This is also supported by Haggard, Howe, and Lynch (Citation2015) who report that big-baths ‘clear the air rather than muddy the water.’ They find substantial evidences of improvements in firms’ information environment, especially in the informativeness of earnings, following big baths.
30. This is supported by prior researches about income smoothing. Tucker and Zarowin (Citation2006) report that income smoothing improves the value of information about future earnings. They also state that stock price contains information about future earnings only in the presence of income smoothing. That is, stock price impounds more information about future performance and cash flows for firms that engage more in income smoothing (Demski Citation1998; Sankar and Subramanyam Citation2001; Kirschenheiter and Melumad Citation2002).