The effect of investor sentiment on the means of earnings management

Prior research has shown that a firm’s tendency to meet or beat earning targets is greater during bad economic times than good times. The paper extends this line of research by investigating which means of earnings management is used in different states of economy. A sample of non-financial companies listed on Korea Securities Market from 2003 to 2011 is used for empirical tests. The findings of this study are summarized as follows. The magnitude of discretionary accruals is negatively related to investment sentiment, indicating that firms tend to use positive discretionary accruals to manipulate reported income upward when the sentiment is pessimistic. However, the real activity based earnings management is not significantly associated with the state of economy. Collectively, this study contributes to behavioral finance and accounting literature by suggesting that managers use discretionary portion of accruals, but do not change their real operating activities, in order to meet or beat earnings targets in economic downturn. Sorah Park (Korea) BUSINESS PERSPECTIVES LLC “СPС “Business Perspectives” Hryhorii Skovoroda lane, 10, Sumy, 40022, Ukraine www.businessperspectives.org The Effect of Investor Sentiment on the Means of Earnings Management Received on: 18th of August, 2017 Accepted on: 6th of December, 2017


INTRODUCTION
There is voluminous accounting research on earnings management, which is defined as "the purposeful intervention in the external financial reporting process with the intent of obtaining some private gain (Schipper, 1989). Prior research (e.g., Graham et al., 2005) has shown that firms tend to inflate earnings to a greater extent during the bad state of economy in order to distinguish themselves from others in the market by boosting stock prices in bad times. In Korea, Park (2015) also documents that firms' tendency to meet or beat earning targets is greater during pessimistic sentiment period.
The literature has detected two major means of earnings management: accrual-based and real-based earnings management. First, many prior studies find that accruals are used at managerial discretion to report more favorable income. For instance, Song et al. (2004) document the empirical results of firms with net income that is slightly below zero or with big loss manipulating earnings upward using accruals, suggesting that accrual-based earnings management is common in Korea. Second, firms may manipulate upward or smooth earnings by involving in abnormal management practices, which are reflected as abnormal operating cash flows, discretionary expenditures and production costs (Roychowdhury, 2006; Kim et al., 2008).
The choice of earnings management appears to vary with firm-specific characteristics and circumstances. For example, Zang (2012) shows that managers adjust the level of accrual manipulation according to the level of real activities manipulation realized, suggesting that managers trade off two methods of earnings management based on their relative costs. Also, Cohen and Zarowin (2010) find that firms' choice between real-and accrual-based earnings management around seasoned equity offerings is a function of firms' ability to use accrual management and the costs of doing so. However, little is known about how equity market sentiment affects the trade-off between real-and accrual-based earnings management. Hence, this paper examines the effect of investor sentiment on accrual-based and real-based earnings management.
The empirical tests of this paper are based on investor sentiment (Baker & Wurgler, 2007) and the level of stock market (Conrad et al., 2002). Using a sample of public companies listed on Korea Securities Market from 2003 to 2011, the paper documents the following empirical test results. First, the magnitude of discretionary accruals is negatively associated with investment sentiment. This implies that firms tend to use positive discretionary accruals to manipulate earnings upward when the sentiment is pessimistic. Second, on the other hand, the real-based earnings management (i.e., operating activities, discretionary R&D expenditures and production activities)is not significantly related to investor sentiment. In sum, these findings suggest that firms do not depend on real activity manipulation, but rather inflate discretionary accruals to report favorable income in bad economic times. Such conclusion is consistent with Graham et al. (2005) arguing that managers are likely to boost earnings in recessions based on their expectation of reversal of intrinsic earnings in economy recovery.
These findings contribute to accounting and behavioral finance research on the relation between macroeconomic condition and financial reporting behavior. Behavioral research presumes that capital market participants including investors and managers are not completely rational in pricing stocks and processing information because they cannot be free from emotions. This study adds new evidence that managers do not manipulate operating activities to boost reported income in bad times, but they tend to rely on discretionary accruals, indicating that accrual quality is affected by sentiment.
The rest of the paper is organized as follows. Section 1 describes the research methods and sample selection. Section 2 reports the empirical test results and last section concludes this study.  Kim and Byun (2010), as in Wurgler (2006, 2007). Wurgler (2006, 2007) combine several imperfect sentiment proxies such as trading volume, divi-dend premium, closed-end fund discount, the number of initial price offerings, the first-day returns on IPOs, and the equity share in new issues to construct the comprehensive measure of investor sentiment. Since only trading volume and equity share in new issues are available and economically significant in Korea, Kim and Byun (2010) form the investor sentiment index by combining four additional variables (i.e., retail investor trading, stock fund flows, customer expectation index, customer's deposit for stock investment).

BSI FUND CEI CD TURN SR
Principal components analysis is performed to isolate the common sentiment components among these six variables from idiosyncratic non-sentiment-related components. Next, KBSENT is the investor sentiment index after controlling for the business cycle. Each of the above six proxies is regressed on the six business-cycle-related variables (i.e., the growth of industrial production, durables sales, semi-durables sales, non-durables sales, service production, and coincident composite index for business cycle changes). Then, the residuals from these regressions are used as the sentiment index controlled for business cycle.
DIFFPE is the estimate of the overall level of equity market following Conrad et al. (2002). It is based on the difference between the market price-to-earnings (P/E) ratio in the current month and the average market P/E over the previous 12 months. Greater DIFFPE indicates that the sentiment in the current month is higher than that in the prior year.

Discretionary accruals
Accrual-based earnings management is measured by the discretionary portion of accruals. The study estimates the discretionary accruals by using the modified Jones model (Dechow et al., 1995;Kothari et al., 2005). The modified Jones model is based on the following regression equation: where, 1 , TAC 2 TAC -total accruals (net income minus operating cash flows, operating income minus operating cash flows) for firm i during year , t scaled by the total assets at the beginning of year , t Rev -sales revenue for firm i in year , t scaled by the total assets at the beginning of year , t PPE -property, plant and equipment for firm i at the end of year , t scaled by the total assets at the beginning of year , t and ROA -return on total assets for firm i in year 1.
t − Then, the discretionary accruals ( ) 1 2 , DA DA are calculated as the difference between total accruals and non-discretionary accruals (i.e., the residuals from the regression equation 1).

Real activity-based earnings management
Following Roychowdhury (2006), the following cross-sectional industry-level annual regressions are used to estimate the normal level of operating cash flows, production costs, and discretionary expenses.
where, CFO -operating cash flows during year , t S -the sales revenue for year , t A -the total assets at the beginning of year , t PROD -the production cost for year ( ) and DISX -selling and general expense-Taxes- The residuals of regression (2), (3) and (4) represent the abnormal , CFOs abnormal production costs, and abnormal discretionary expenses, respectively. Finally, real-based earnings management measures are defined as follows for simplicity:

Empirical test model
The following regression model is estimated to examine the research question:  The sample period ends in 2011 in which Korean International Financial Reporting Standard (K-IFRS) was adopted. Also, I delete firms with non-December fiscal year-end, impaired capital, negative total assets or negative book equity in order to ensure comparability in the sample. The final sample contains 10,723 firm-year observations (489 distinct firms).   , , KSENT KBSENT DIFFPE are negatively associated with discretionary accruals. Also, these negative correlations are statistically significant at the 1% level. This indicates that firms tend to manipulate earnings upward using positive discretionary accruals when the investor sentiment is pessimistic.  Collectively, these findings suggest that firms do not appear to rely on changing their real operating activities such as timing of sales, R&D and production to report favorable earnings. Rather, firms tend to use discretionary accruals to boost reported income in bad times. Hence, the means of earnings management appear to vary with investor sentiment.  Note: All t-values are based on two-tailed t-tests. ***, **, and * indicate the statistical significance at 1%, 5%, and 10% levels, respectively.

CONCLUSION
Using a sample of public companies on Korea Securities Market from 2003 to 2011, this paper show that firms do not rely on real activity manipulation rather they tend to inflate earnings using discretionary accruals in bad economic times. Consistent with Graham et al. (2005), managers tend to use positive discretionary accruals in recessions because intrinsic earnings will increase when the economy recovers which leads to the reversal or catch-up.  Note: All t-values are based on two-tailed t-tests. ***, **, and * indicate the statistical significance at 1%, 5%, and 10% levels, respectively.