University of Amsterdam Faculty of Economics and Business
BSc. Economics and Business
Bachelor Thesis
Earnings management activities during the ongoing European Sovereign Debt Crisis
Malina Ioana Stanciu
Student number: 10232974 Supervised by: Dr. Dirk Damsma Date: June 2015
1 Statement of Originality
This document is written by Malina Ioana Stanciu, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2 Table of contents
I. Introduction ... 3
II. Theoretical framework ... 5
2.1 Financial and accounting literature ... 5
2.2 Macro-economic and political implications ... 7
III. Methodology and descriptive statistics ... 9
3.1 Data and sample description ... 9
3.2. Econometric model and dependent variables ... 10
3.3. Independent variables ... 12
IV. Analysis and results ... 16
4.2. Expectations, descriptive statistics and correlation ... 17
4.3. Regression results ... 18
3 Earnings management activities during the ongoing European Sovereign Debt Crisis
Abstract
This study investigates whether earnings management activities have a directly proportional relationship with the intensity of the European Sovereign Debt Crisis. That is they increase when the intensity of the crisis is high, and they decrease as the intensity of the crisis decreases. The analysis is focused on small, mid and large capitalization companies from the Euro Zone. As a measure for earnings management activities, this study uses discretionary accruals. The intensity of the crisis is measured using the Composite Index of Systemic Stress. The findings show no conclusive evidence that earnings management activities increase or decrease in accordance with the intensity of the European Debt Crisis.
I. Introduction
The magnitude of the 2008 economic crisis that began in the United States, spread to Europe, and eventually became global, calls for a structural reassessment of business, economics and especially accounting research (Arnold, 2009). Existing literature in the field debates the faultiness of financial reporting rules and fair value accounting rules (eg. Magnan, 2009; Iatridis & Dimitras, 2013), and suggests that these are the most obvious ways through which the world of accounting practice is involved in the current economic crisis (Arnold, 2009). Furthermore, as the crisis moved to Europe both directly and indirectly through emerging markets, it disclosed a series of fundamental imbalances in the way European economies have been operating so far, leading the European economy into a severe recession (Watt, 2008). As such, this paper attempts to build upon existing literature in developing a better understanding of the relationship between accounting practice and the macro political and economic environment in which it operates.
Both Arnold (2009) and Iatridis & Dimitras (2013) emphasize in their work the importance of rigorous corporate disclosure in light of a financial crisis, for accounting regulators and investors. They point out the value relevance of reported financial numbers and state that while investors aim to establish a profitable investment strategy, accounting regulators aim to increase the quality of reported disclosures, reduce information asymmetry and diminish earnings manipulation behavior (Arnold, 2009). At the same time, Lane (2012) infers that the
4
current European Sovereign Debt Crisis is highly related to the banking crisis in the United States and has as an additional complexity the macroeconomic imbalances residing within the Euro Area. As such, this debt crisis has also led to a crisis of confidence for European businesses and economies (European Sovereign Debt Crisis, 2015). In their study, Gorgan et al (2012) state that confidence is what financial markets rely on the most. What supports this confidence is the assumption that financial statements offer a clear and fair view of an institution’s financial position, performance and changes in equity. However, the integrity of accounting information offered to investors and other stakeholders has been challenged under the circumstances of the crisis. Issues such as information asymmetries - that reflect the degree to which managers have more information about the firm than other contracting parties - and agency conflicts between management and stakeholders can lead to earnings management situations (Dechow, 1994).
Degeorge, Patel, & Zeckhauser (1999) define earnings management as “the strategic exercise of managerial discretion in influencing the earnings figure reported to external audiences”. Earnings ultimately drive stock prices. An increase in earnings usually results in stock prices moving up and vice versa. If earnings do not reliably portray a company’s financial situation, its stock price might mislead investors regarding the future profitability of their investment. This may cause a loss of investor confidence in the market and the economy would suffer consequently, as Gorgan, Gorgan, Dumitru, & Pitulice (2012) suggest has happened since 2008. This view is enforced by Coffee (2009) who introduces the concept of a “gatekeeper” in the market. He defines gatekeepers as “reputational intermediaries who provide verification and certification services to investors, doing essentially what investors cannot easily do for themselves”. According to Coffee (2009), during the period preceding the 2008 market collapse, gatekeepers such as auditors, security analysts, investment bankers and especially credit rating agencies failed to reliably and objectively monitor corporate managements. He then infers that the 2008 market bubble burst when investors recognized a systematic failure of the gatekeepers, and points out that the crisis was “a by-product of excessive reliance on a gatekeeper, which became increasingly subject to client pressure as competition increased in its market” (Coffee, 2009).
On another note, Trombetta & Imperatore (2014) infer that during a period of financial turmoil managers have to address firm survival as a more critical issue than in normal times. This implies that in order to achieve this goal, managers might engage in earnings management
5
activities, according to the severity of the financial crisis. This means that when the financial turmoil is very serious and it is severely affecting companies, earnings management activities tend to spike as managers are struggling to keep the companies at float. As the financial turmoil decreases in severity, but the economy has not yet recovered to its initial pre-crisis state, earnings management activities are still persistent, but to a lesser extent than in the previous scenario. As such, the reminder of this paper sets out to test the extent to which the intensity of the crisis in Europe influences earnings management activities. That is, this paper intends to examine whether during periods of intense financial instability more earnings management practices can be observed, and during periods of less financial instability, less earnings management practices can be observed.
In the next sections of this paper, both a theoretical framework that describes the research conducted so far on this topic and the process through which this paper intends to address the issue will be developed. Furthermore, together with a statement of the initial hypothesis, a description of the data and the methodology used will be presented. Finally, an analysis of the data and its results will conclude the paper.
II. Theoretical framework
In answering the question of whether more earnings management practices were observed during high intensity periods of the European Sovereign Debt crisis and less earnings management practices were observed during low intensity periods of the crisis, within the Euro Area, previous literature on the topic has to be taken into account. This is in order to better devise a methodology consistent with previous studies and build upon potential gaps in the research conducted so far. Firstly, some more insight into the underlying details of earnings management and its importance will be presented. Lastly, in the other half of this chapter, a presentation of the link between periods of financial distress and earnings management activities will be compiled.
2.1 Financial and accounting literature
A company’s earnings are the revenue from its economic activity, minus all the costs involved. They are the result of the application of accrual rules and approximations of anticipated long term cash flows at a certain point in time. Even though in the long run the stream of earnings and
6
the stream of cash flows will converge, managers have the space within accounting earnings recognition rules and outside them to bend the timing of the recognition of earnings. Stolowy, Lebas, & Ding (2013) argue that such “account manipulation” practices are based on a common functional view that accounting numbers contribute to the determination of share prices. However, as the recognition of accruals is usually at the discretion of the management of a company, this might create opportunities for earnings manipulation, that is situations in which managers’ choice of reporting methods and estimates do not accurately portray their firm’s financial information (Healy & Wahlen, 1999).
In an attempt to converge and harmonize this process, on the 19th of July 2002 the European Parliament approved Regulation (EC) No 1606/2002 of the European Parliament and of the Council of 19 July 2002 on the application of international accounting standards. As a result, all European Union listed companies are required to prepare their financial statements following the International Financial Reporting Standards as of 2005. Nevertheless, there is a question of whether situations of accrual based earnings management still happened after 2005 and moreover starting in 2008, when the financial crisis started and IFRS reporting guidelines have already been in place for three years. As Iatridis & Dimitras (2013) suggested, factors impacting the decision of earnings manipulation can be: personal goals of managers, pressures on the firm made by external parties, corporate culture or a need to face firms’ obligations. Irrespective of the reason, the actions that managers are prone to take in this direction could be divided into two categories: cash based earnings management (or real activities manipulation), and accrual based earnings management (Zang, 2012).
Stolowy, Lebas, & Ding (2013) state that between the two forms of earnings management – cash based and accrual based – the one that has been more extensively discussed in the literature is related to managements’ decision on the level of accruals recorded in the financial statements; that is on the accuracy of a company’s earnings estimates. They continue to argue that the cause for low quality earnings can be found on the path used by accountants to progress from income to cash flow from operations. That is through groups of adjustments made, such as non-cash items, gain or loss on sale of fixed assets and changes in inventories, accounts receivable and accounts payable. For instance, a company might report a high net income, but a low cash flow from operating activities, which might be due to a possible increase in inventories or accounts receivable, and an unusual decrease in payables. This can be because of a company’s
7
mismanagement of its working capital needs or because of a manipulation of its earnings by misleadingly accruing for anticipated sales and artificially reducing its cost of goods sold.
As a result, in an attempt to better determine a cause of low quality earnings and to more accurately isolate earnings management activities from a pure mismanagement of a company’s resources, Dechow, Sloan, & Sweeney (1995) evaluate several accrual based models for detecting earnings management. Their study is concluded by stating that the model which shows the most power in detecting earnings management is a modified version of the model developed by Jones (1991). The modified version of the Jones model splits accruals intro two categories, discretionary and non-discretionary, and uses the first category to measure earnings management activities. The model functions on the assumption that “it is easier to manage earnings by exercising discretion over the recognition of revenue on credit sales than it is to manage earnings by exercising discretion over the recognition of revenue on cash sales” (Dechow, Sloan, & Sweeney, 1995). As such, discretionary accruals as computed by the Modified Jones model became a widely used proxy to measure earnings management activities by researchers in the field, and this paper will follow in their footsteps. A more detailed explanation about how exactly does the Jones Model work will follow in the third chapter of this paper.
2.2 Macro-economic and political implications
Previous research in the field has also investigated how managers’ accounting choices are affected by a period of financial distress. On the one hand, it has been inferred that periods of financial distress affect accounting quality and implicitly the accuracy of the reported earnings; that is, financial instability affects manager’s accounting choices. Trombetta and Imperatore (2014) suggest that managers may engage in earnings management activities in order to affect the market’s evaluation of a firms’ probability to survive the financial turmoil in which it operates, and hence reduce its average cost of capital. This will attract investors, and the company will satisfy its liquidity requirements in order to continue its activity. Choi, Kim, & Lee (2011) concluded in their study that “the Asian financial crisis of 1997-1998 led to a significant decline in the information value of discretionary accruals”, and have indeed noticed “an increased use of opportunistic earnings management during the crisis”. Furthermore, Kousenidisa, Ladas, and Negakis (2013) affirmed in their study that firms that have larger absolute discretionary accruals in the crisis period were found to have lower quality earnings. As
8
such, they concluded that when incentives for earnings management are high, firms were found to have less predictable, lower quality earnings.
On the other hand, Kousenidisa, Ladas, and Negakis (2013) study results also suggest that during the period of the crisis, firms appear to report “earnings that are timelier, more conditionally conservative, more value relevant, less smoothed, less managed, more persistent and more predictable.” Adding up to this view, Filip & Raffournier (2014) found that earnings management activities of European-listed firms during the 2008-2009 financial crisis have significantly decreased; confirming this for most of the 16 European countries under review, but not for all. This leads us to point out that notably lacking from the existent literature on earnings management activities during the European Sovereign Debt crisis is the examination of the relationship between the intensity of a financial crisis and earnings manipulation behaviors. Trombetta & Imperatore (2014) have addressed this issue in their research and concluded that the association between earnings management and financial crises is non-monotonic. That is, earnings quality improves during moderate periods of the crisis and worsens as the financial crisis becomes extreme. They justify their stance by arguing that managers are less willing to engage in risky activities that might threaten the survival of the company, when operating in an already financially instable economy; but as the instability increases, managers are more likely to take risks and do everything in their power to keep the company afloat.
However, their research has been focused on a single country setting, rather than a setting with a wider variety of characteristics in terms of culture, the level of economic development and the level of financial development, such as the Euro Zone; a fact to be addressed in this paper. Moreover, The European Sovereign Debt Crisis provides a unique opportunity to study earnings management within the Euro Zone, as it is the first crisis to hit this affected area since the Economic and Monetary Union (EMU) was ratified in Europe both in terms of duration and magnitude (Kousenidisa, Ladas, & Negakis, 2013). It is also a different type of crisis which according to Lane (2012) started as a financial crisis and continued to become a full-blown debt crisis with a strong fiscal component. The study of the relationship between earnings management activities and the intensity of such a distinctive type of crisis builds upon current research in the field and might provide additional insight into the topic.
To conclude, current research on the topic describes the financial incentives and space within accounting rules that a company’s management has in order to engage in earnings
9
management activities. Furthermore, the macro-economic instability that can be observed during a global financial crisis, and the implications this turmoil has for companies, is suggested to give managers an additional challenge to surpass. Hence, taking all the above into account, the main hypothesis to be tested in this paper is: During periods of the European Sovereign Debt crisis of
a lower intensity, earnings management activities as proxied by discretionary accruals are observed to be low; as the intensity of the crisis increases, more earnings management activities can be observed. The next chapter will present how the testing will proceed, followed by a
presentation of the results and some concluding remarks.
III. Methodology and descriptive statistics
In accordance with the reviewed literature in the previous section, an analysis of the hypothesis as to whether earnings management activities increase or decrease with the intensity of the financial crisis; will be conducted as follows. Firstly, the data and sample description will be introduced, followed by a presentation of the metrics used to compile the regression model. Secondly, some descriptive statistics of these metrics will be portrayed.
3.1 Data and sample description
The analysis was conducted on a sample of companies listed in the EURO STOXX® ex Financials Index. This index represents large, mid and small capitalization companies from 12 Eurozone countries: Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal and Spain, and excludes stocks that are allocated to the industry 'Financials' according to the Industry Classification Benchmark (STOXX Indices, 2015). Yearly data was collected using the Datastream database for the period between 2007 and 2014. As the years 2007 and 2014 had incomplete and insufficient data to be included in the analysis, they were excluded completely. Moreover, out of the 227 companies compiling the index only 192 had the complete data required for the analysis. As such, 192 companies with 6 years of data each, gives us an approximate number of 1,156 observations of yearly data per company to be analyzed, spread across the 12 countries covered in the index, as presented in Table I. The uneven spread of the data amongst countries is derived mainly from the uneven spread of the companies in the index amongst their country of origin. That is, in the Euro Zone,
10
France and Germany for instance host more large, mid and small capitalization companies than Austria or Portugal.
Country Freq. Percent Cum.
AT 19 1.64 1.64 BE 42 3.63 5.28 DE 276 23.88 29.15 ES 95 8.22 37.37 FI 92 7.96 45.33 FR 356 30.8 76.12 GR 12 1.04 77.16 IE 44 3.81 80.97 IT 96 8.3 89.27 LU 12 1.04 90.31 NL 95 8.22 98.53 PT 17 1.47 100 Total 1,156 100
Table I: Spread of observations of yearly company data
3.2. Econometric model and dependent variables
The regression model of this paper was compiled using 2 categories of metrics: earnings management metrics and financial crisis metrics. As a proxy for the magnitude of earnings management activities, discretionary accruals were computed for each of the 192 firms in the sample individually; that is for each grouping of the panel data used. They were computed using the cross-sectional version of the Modified Jones model, as described in Trombetta & Imperatore (2014). The modified version of the Jones model is in essence a regression of the total accruals of a company on several independent variables such as the change in current assets, change in revenue and change in a company’s property, plant and equipment. The output of this regression, which is the resulting parameter estimates, is then used to run a second regression and calculate the non-discretionary component of total accruals. Lastly, the discretionary component is calculated by a simple decrease of the non-discretionary accruals amount from total accruals. A more detailed explanation of the model is presented as follows.
Firstly, total accruals lagged by the amount of total assets at the beginning of the year were calculated for each firm in the sample using the following regression model:
𝑇𝐴𝑖𝑡 𝐴𝑖(𝑡−1) = 𝛼0+ 𝛼1( 1 𝐴𝑖(𝑡−1)) + 𝛼2( ∆𝑅𝐸𝑉𝑖𝑡 𝐴𝑖(𝑡−1)) + 𝛼3( 𝑃𝑃𝐸𝑖𝑡 𝐴𝑖(𝑡−1)) + 𝜀𝑖𝑡
11
where 𝐴𝑇𝐴𝑖𝑡
𝑖(𝑡−1) = firm i’s total accruals in year t divided by total assets at the beginning of the year;
that is at the end of the previous year. Firm i’s total accruals in year t were calculated as follows: 𝑇𝐴𝑖𝑡 = ∆𝐶𝐴𝑖 − ∆𝐶𝑎𝑠ℎ𝑖 + ∆𝐶𝐿𝑖− ∆𝑆𝑇𝐷𝑖 − 𝐷𝑒𝑝𝑟𝑖.
where ∆𝐶𝐴𝑖 = firm i’s change in current assets from year t-1 to year t, ∆𝐶𝑎𝑠ℎ𝑖 = firm i’s change in cash and short term securities from year t-1 to year t, ∆𝐶𝐿𝑖 = firm i’s change in current liabilities from year t-1 to year t, ∆𝑆𝑇𝐷𝑖 = firm i’s change in short term debt from year t-1 to year t and 𝐷𝑒𝑝𝑟𝑖 = firm i’s depreciation, amortization and depletion amount in year t. Moving on to the next variables, ∆𝑅𝐸𝑉𝑖𝑡
𝐴𝑖(𝑡−1) = firm i’s change in revenue from year t-1 to year t divided by total
assets at the beginning of the year and 𝑃𝑃𝐸𝑖𝑡
𝐴𝑖(𝑡−1) = firm i’s property plant and equipment amount
divided by total assets at the beginning of the year. Secondly, the non-discretionary part of the total accruals for firm i in year t was calculated using the 𝛼0, 𝛼1, 𝛼2, 𝛼3 parameter estimates calculated in the regression above, as indicated in the below equation:
𝑁𝐴𝑖𝑡 = 𝛼0+ 𝛼1( 1 𝐴𝑖(𝑡−1)) + 𝛼2( ∆𝑅𝐸𝑉𝑖𝑡 − ∆𝑅𝐸𝐶𝑖𝑡 𝐴𝑖(𝑡−1) ) + 𝛼3( 𝑃𝑃𝐸𝑖𝑡 𝐴𝑖(𝑡−1))
where ∆𝑅𝐸𝐶𝑖𝑡 = firm i’s change in account receivables from year t-1 to year t. Lastly, the measure of discretionally accruals needed was computed as the difference between total accruals lagged by total assets and nondiscretionary accruals, defined as:
𝐷𝐴𝑖𝑡 = ( 𝑇𝐴𝑖𝑡
𝐴𝑖(𝑡−1)) − 𝑁𝐴𝑖𝑡
Similar to previous research in the field, this paper used three variants of discretionary accruals as dependent variables of the regression model, in order to better identify the magnitude and the potential intent of managers when engaging in earnings management activities. Absolute discretionary accruals (ABS_DA) are analyzed to pinpoint the magnitude of earnings management situations, whereas negative discretionary accruals (Neg_DA) and positive discretionary accruals (Pos_DA) were used to pinpoint income decreasing and income increasing earnings management respectively (Trombetta & Imperatore, 2014). As such, the regression model will have three variants, one for each dependent variable studied.
12 3.3.Independent variables
In terms of independent variables of the model, this paper will use the ones summarized in Table II below, and presented as follows. The dependent variables of this paper were chosen such as several characteristics of the sample studied were controlled for: macroeconomic environment characteristics, business cycle characteristics, country characteristics and firm level characteristics. For the purpose of our research, characteristics of the macroeconomic environment and characteristics of the business cycle are the main independent variables (Trombetta & Imperatore, 2014), whereas firm and country variables are control variables.
Firstly, as a variable used to measure macroeconomic characteristics of the economy, a financial stress index was used. The main overall purpose of using stress indices is to measure the current state of instability in the financial system and to comprise it into a single statistic (Hollo, Kremer, & Lo Duca, 2012). In their research, Trombetta & Imperatore (2014) used the Kansas City Financial Stress Index (KCFSI) proposed by Hakkio and Keeton (2009), and stated that previous academic literature on the topic adopted several other approaches to measure the intensity of financial crises. Examples of such approaches are: relying on historical narratives (that is using quantitative thresholds for the chronological events of well-known financial crises) or using binary variables of underlying financial variables. However, due to the specific characteristics of the Eurozone countries sampled (level of development of the market, regulation, culture and others), a financial stress index introduced by economists of the European Central Bank applicable to the Eurozone was used in this paper. This index is the Composite Indicator of Systemic Stress in the financial system (CISS) developed by Hollo, Kremer, & Lo Duca (2012).
ECB researchers and developers of the CISS define the specific aim of the CISS to be the one of highlighting the systemic nature of the existing turmoil in the financial system, where systemic stress can be interpreted as a risk which has already materialized (Hollo, Kremer, & Lo Duca, 2012). The CISS measures current level of stress in several segments of an economy’s financial system on the basis of 3 stress indicators capturing some specific symptoms of financial distress such as: increases in agents’ uncertainty, investor disagreement or information asymmetries (Hakkio & Keeton, 2009). As such, taking into account the time frame of the ongoing European Debt Crisis, the Euro Area CISS appears to be peaking during well-known periods of elevated financial distress, specifically the years 2008 and 2009 as indicated in Figure
13
I, and it decreases as the economy is thought to be recovering slowly. In addition, a yearly squared value of the CISS was used in the model in order to avoid multicollinearity issues.
Fig I: CISS Euro Area for the period of 2007 – 2014
Secondly, as a variable used to measure business cycle characteristics, the Economic Sentiment Indicator (ESI) was used. According to the European Commission (2015), periodic surveys allowing comparisons among different countries' business cycles are conducted for different sectors of the economies in the European Union (EU), and have become an essential tool for monitoring the evolution of the EU and the Euro area economies. The European Commission (2015) defines confidence indicators as arithmetic means of seasonally adjusted balances of answers to a selection of questions closely related to the reference variable they are supposed to track. The ESI is a composite indicator that reflects and represents judgments and attitudes of producers and consumers. It made up of five sectorial confidence indicators with different weights as follows: Industrial confidence indicator [40%]; service confidence indicator [30%]; consumer confidence indicator [20%]; construction confidence indicator [5%]; retail trade confidence indicator [5%]. The economic sentiment indicator (ESI) is calculated as an index with mean value of 100 and standard deviation of 10 over a fixed standardized sample period (Eurostat, 2015). Figure II illustrates the trend taken by the economic sentiment of the Euro Zone, as measured by the ESI index, over the sample period.
2007 2008 2009 2010 2011 2012 2013 2014 Series1 0.17741 0.52344 0.55844 0.30478 0.32721 0.31413 0.07103 0.09117 0.00 0.10 0.20 0.30 0.40 0.50 0.60 Sc al e o f 0 t o 1 Years Series1 Trendline
14 Fig II: Annual ESI Euro Area for the period of 2007 – 2014
Thirdly, country characteristics were controlled for through 2 variables: the level of economic development proxied by the GDP per capita of each country sampled and the level of economic development proxied by stock market capitalization. Gaio (2010), argues in her research that firms in countries with poor economic development may have fewer incentives to assure better earnings quality, as credible external verification of accounting information may be too costly for them. In this paper, we measure the level of economic development (Log(GDP)) as the logarithm of gross domestic product per capita, measured in current international dollars, following Gaio’s (2010) methodology. Furthermore, Gaio (2010) also infers that in countries with poorly developed stock markets, access to external funds is limited to firms and as such, it is harder and more costly for them to raise external funds and consequently impose on themselves better earnings quality. In this paper, the level of financial development (MktCap/GDP) is proxied by stock market capitalization divided by gross domestic product also following Gaio’s (2010) methodology.
Finally, firm level characteristics were controlled for through several variables, in particular: length of the operating cycle (Lgth_OPCYLE), firm size (Log(TAssets)), firm leverage (Lev), firm performance (ROA and CFO), all presented as follows. First of all, longer operating cycles indicate more uncertainty in the operating environment; that is, longer operating cycles imply longer accounts receivables days and a longer timespan between the product has been sold or the service rendered and the time to collect the revenue for it. Moreover, the risk of not collecting the revenue at all increases as the operating cycle is longer and together with it,
2007 2008 2009 2010 2011 2012 2013 2014 Annual ESI Euro Area 110.6 94.4 80.3 101.3 102.2 90.6 93.6 101.3
0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 Sc al e Years
Annual ESI Euro Area Trendline
15
there is more uncertainty. This uncertainty implies manager judgment is more often required, so there is a higher probability of error estimation and poorer accruals quality (Gaio, 2010). As such, this paper uses as a proxy for the length of the operating cycle the logarithm of the sum of days inventories divided by the accounts receivables days.
Furthermore, firm size is also considered in the research model as it is expected that firms of a larger size are more carefully monitored by the market, auditor firms and other stakeholders, which makes engaging in earnings management activities more difficult (Trombetta & Imperatore, 2014). As a proxy for firm size, this paper uses the logarithm of a firm’s total assets. The next variable used in our model is the leverage of a firm, proxied by total debt divided by total assets. This was incorporated in the model as previous research shows that firms facing financial constraints have an incentive to manipulate earnings in order to avoid disclosing their real financial situation and bear potential losses as a result (Trombetta & Imperatore, 2014). A variable to control for firm performance was also introduced in the model. This was proxied by two measures of a company’s performance that according to Trombetta & Imperatore (2014), represent potential determinants of earnings management: the return on assets and the cash flow from operating activities. Furthermore, Trombetta & Imperatore (2014) also implied in their research that earnings management are particularly sensitive to the firm’s operating volatility like volatility of sales and volatility of cash flows, and that the variable proxying the financial crisis is likely to be correlated to the two. Hence, in order not to create an omitted variable bias in the model, we will use the standard deviation of sales as proxy for the volatility of sales and the standard deviation of cash flows as a proxy for the volatility of cash flows. A summary of the variables and their description is presented in Table II below.
To conclude, the model used to test the hypothesis of this paper is a regression model where the dependent variable is discretionary accruals as calculated using the Modified Jones model and who act as a proxy for measuring earnings management activities. Also, the main independent variables are the financial crisis’s indicator together with real business cycle, and all the other country and firm level variables are control variables (Trombetta & Imperatore, 2014).
𝐴𝐵𝑆_𝐷𝐴𝑖𝑡 = 𝛽0+ 𝛽1𝐶𝐼𝑆𝑆𝑡+ 𝛽2𝐶𝐼𝑆𝑆𝑡2+ 𝛽
3𝐸𝑆𝐼𝑡+ 𝛽4𝐿𝑜𝑔(𝐺𝐷𝑃)𝑡+ 𝛽5𝑀𝑘𝑡𝐶𝑎𝑝/𝐺𝐷𝑃𝑡
+ 𝛽6𝐿𝑔𝑡ℎ_𝑂𝑃𝐶𝑌𝐶𝐿𝐸𝑖𝑡+ 𝛽7𝐿𝑜𝑔(𝑇𝐴𝑠𝑠𝑒𝑡𝑠)𝑖𝑡+ 𝛽8𝐿𝑒𝑣𝑖𝑡+ 𝛽9𝑅𝑂𝐴𝑖𝑡+ 𝛽10𝐶𝐹𝑂𝑖𝑡 + 𝛽11𝑆𝐷𝑅𝑒𝑣𝑖𝑡+ 𝛽12𝑆𝐷𝐶𝐹𝑂𝑖𝑡+ 𝜀𝑖𝑡
16
Label Description and measurement
Dependent variable
ABS_DA Absolute value of a firm’s discretionary accruals Pos_DA Positive values of discretionary accruals
Neg_DA Negative Values of discretionary accruals
Independent variables
CISS Monthly value of the Eurozone Composite Indicator of Financial Stress index averaged to a year total
CISSsq Squared year total of the Eurozone Composite Indicator of Financial Stress index
ESI Monthly value of the Eurozone Economic Sentiment Indicator averaged to a year total
Control Variables
Log(GDP) Logarithm of a country’s yearly GDP per Capita, priced in current international dollars
MktCap/GDP Yearly market capitalization of a firm divided by the firm’s country yearly GDP per Capita, priced in current international dollars
Lgth_OPCYCLE Yearly operating cycle length of a firm, calculated as the logarithm of sum of days inventory divided by days accounts receivable
Log(TAssets) Logarithm of a firm’s yearly total assets
Lev Yearly leverage of a firm, calculated by total debt divided by total assets ROA Yearly return on assets of a firm, extracted from Datastream
CFORev Yearly cash flow from operations of a firm, calculated by cash flows from operations divided by total revenue/total sales
SDRev Volatility of revenue/sales of a firm, calculated as the standard deviation of of revenue in previous 4 years
SDCFO Volatility of cash flow from operations of a firm, calculated as the standard deviation of cash flow from operations in previous 4 years
Table II: Variables description and measurement
IV. Analysis and results
In accordance with the methodological framework in the previous section, an analysis of the hypothesis as to whether earnings management activities increase or decrease with the intensity of the financial crisis was conducted, and the results will be presented as follows. Firstly, the expectations this study has regarding the econometric model used will be presented. Secondly, some descriptive statistics and correlation of the variables will be introduced, and lastly, the results of the regression analysis and their interpretation will be portrayed.
17 4.2. Expectations, descriptive statistics and correlation
Our expectations regarding the econometric model presented above are mostly related to the CISS index variable, but also to the ESI index variable. We expect this model to provide conclusive evidence that the intensity of the financial crisis does influence earnings management activities in either a positive or a negative way, according to the level of the CISS index. That is, we expect an increase in earnings management activities as the CISS index increases, and a decrease otherwise. When we turn to the ESI index, we expect the opposite. Our theory is that as the ESI index increases, earnings management activities will decrease. These expectations are illustrated by Figure III below.
Fig III plots the trend of the CISS index and the periods of low confidence in the market as indicated by the Economic Sentiment indicator within the period of 1999 – 2015, the timespan since the establishment of the Euro Zone in 1999. It can be clearly observed that the relationship between the two is inversely proportional. That is, when the financial crisis peaks in intensity, during the 2008 - 2009 period, the ESI indicates a sharp decrease of consumer and producer confidence in the market, whereas outside this period, the opposite can be observed.
Fig III: Comparison of financial crisis and business cycle effects over the period 1999 – 2015 0 0.1 0.2 0.3 0.4 0.5 0.6 20.00 40.00 60.00 80.00 100.00 120.00 140.00 Jan -99 Oct-99 Ju l-00 Ap r-01 Jan -02 Oct-02 Ju l-03 Ap r-04 Jan -05 Oct-05 Ju l-06 Ap r-07 Jan -08 Oct-08 Ju l-09 Ap r-10 Jan -11 Oc t-11 Ju l-12 Ap r-13 Jan -14 Oct-14 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 ESI 101 101 93. 90. 102 101 80. 94. 110 108 100 101 95. 96. 103 116 107 CISS 0.11 0.09 0.07 0.31 0.33 0.3 0.56 0.52 0.18 0.09 0.09 0.07 0.11 0.18 0.16 0.12 0.08 ESI CISS
18
The above trends are also consistent with the descriptive statistics presented in Table III. More precisely, the mean value of the CISS over the sample period under consideration is above the average of the 1999 – 2015 timespan, which indicates that indeed during 2007- 2014 a period of intense financial distress has been recorded. Table IV presents the correlation matrix for all the variables. Noteworthy is that the CISS is not significantly correlated with the dependent variable, ABS_DA.
Table III: Descriptive statistics for the full sample, for a total number of 1,156 observations
Table IV: Correlation matrix
4.3. Regression results
We begin our analysis by investigating whether there is a statistically significant relationship between earnings management and the intensity of the financial crisis in the Euro Zone. Our first
Variable Mean Std. Dev. Min Max
ABS_DA 0.04708 0.13097 0.00018 2.83716 CISS 0.34698 0.16229 0.07103 0.55844 ESI 93.96773 7.52893 80.29166 110.62500 Log(GDP) 4.58231 0.06256 4.40804 4.95622 MktCap/GDP 28.64692 18.45343 0.74891 81.48434 Lgth_OPCYCLE 1.21844 1.50154 (0.80989) 11.32759 Log(Tassets) 7.01959 0.60170 4.76112 8.42384 Lev 0.28687 0.15163 - 1.16535 ROA 6.10631 6.38374 (35.92000) 69.32000 CFO 0.13583 0.11055 (0.19420) 0.97115 SDRev 1,920,129 3,730,282 10,980 38,200,000 SDCFO 447,721 728,374 5,473 6,229,920
ABS_DA CISS ESI LogGDP MktCap~P Lgth_O~E LogTAss Lev ROA CFORev SDRev SDCFO ABS_DA 1.00000 CISS 0.04070 1.00000 ESI 0.00380 (0.46140) 1.00000 LogGDP (0.09660) (0.17480) 0.08040 1.00000 MktCapGDP (0.03400) (0.02150) (0.02260) (0.30000) 1.00000 Lgth_OPCYCLE (0.02640) (0.01230) 0.01480 (0.01740) 0.02860 1.00000 LogTAss 0.00080 (0.06270) 0.01970 (0.06140) 0.15120 (0.09700) 1.00000 Lev (0.10490) 0.05690 (0.03950) (0.14680) (0.06400) (0.19380) 0.18320 1.00000 ROA (0.00970) (0.02360) 0.07730 (0.01020) (0.11840) 0.09460 (0.36350) (0.21830) 1.00000 CFORev (0.05950) 0.04660 (0.05590) (0.00420) (0.05150) (0.05400) (0.06350) 0.32690 0.21180 1.00000 SDRev 0.04080 0.00220 (0.01390) (0.02920) 0.05700 (0.07980) 0.52240 0.00460 (0.11150) (0.13760) 1.00000 SDCFO 0.03240 (0.00650) 0.01790 (0.05160) 0.06720 (0.09410) 0.62320 0.07030 (0.16680) (0.10680) 0.72430 1.00000
19
regression model is using the absolute value of discretionary accruals to measure earnings management activities. The second and the third model is using positive discretionary accruals and respectively, negative discretionary accruals. The estimation results of these three models using robust standard errors and significance levels of 1%, 5% and 10% are presented in Table V, VI and VII below.
ABS_DAit= β0+ β1CISSt+ β2CISSt2+ β
3ESIt+ β4Log(GDP)t+ β5MktCap/GDPt
+ β6Lgth_OPCYCLEit+ β7Log(TAssets)it+ β8Levit+ β9ROAit+ β10CFOit
+ β11SDRevit+ β12SDCFOit+ εit
Table III : Model 1 - Panel regression model of earnings management on financial crisis metrics and firm level variables
20
The results for all three models are insignificant at significance levels of 1%, 5% and 10%. The coefficient for the CISS is negative and insignificant at significance levels of 1%, 5% and 10%, for all three models. This means that financial stress on the economy does not seem to have a significant effect on earnings management activities. Furthermore, the negative sign of the coefficient indicates an overall low level of earnings management activities during a period of financial turmoil. However, the p-value for the model where positive discretionary accruals are used as a proxy for earnings management activities is smaller than in the other models, and closer to at least the threshold of the 10% significance level. This might hint towards a trend of more income increasing earnings management activities during a period of financial turmoil, which would be consistent with our expectations.
When turning to analyze the impact macroeconomic factors have on the level of earnings management activities, we observe the ESI variable to have statistically significant coefficients at significance levels of 10% and 5% respectively, in the models where discretionary accruals have either an income increasing or income decreasing direction - that is, in the second and third model. These results provide evidence that the economic sentiment in the economy is indeed a factor that influences the level of earnings management activities. However, the positive sign of the coefficient suggests that the level of earnings management activities moves together with the economic sentiment indicator. That is, if during a period of financial distress there is a loss of confidence in the market, earnings management activities will decrease. This is not consistent with our expectations.
The control variables used in the three models have given in general insignificant results at significance levels of 1%, 5% and 10%. This implies that neither country nor firm level characteristics have a significant impact on the level of earnings management activities in the economy. However, the coefficients of some variables have proven to be significant, that is to have an impact in on earnings management activities. These variables are the LogTAss and the CFORev. The natural logarithm of Total Assets has a significant coefficient at a 1%, 5% and 10% significance level in the third model. This means that the size of the company, proxied by the natural logarithm of total assets does impact positively income decreasing earnings management activities. That is as the size of the company grows bigger, the level of discretionary accruals decreases. This is in line with what we expect, and with previous studies conducted (Iatridis & Dimitras, 2013), as larger companies tend to have regular audits in place
21
conducted by one of the big 4 auditing firms. The coefficient of the CFORev control variable is also significant at a significance level of 10%, and has a negative sign, in the second model. That is as firm performance increases, a decreased level of discretionary accruals can be observed. This result is also consistent with our expectations since companies performing well do not feel the need to engage in earnings management activities.
Pos_DAit= β0+ β1CISSt+ β2CISS2t + β3ESIt+ β4Log(GDP)t+ β5MktCap/GDPt
+ β6Lgth_OPCYCLEit+ β7Log(TAssets)it+ β8Levit+ β9ROAit+ β10CFOit
+ β11SDRevit+ β12SDCFOit+ εit
22 Neg_DAit= β0+ β1CISSt+ β2CISSt2+ β
3ESIt+ β4Log(GDP)t+ β5MktCap/GDPt
+ β6Lgth_OPCYCLEit+ β7Log(TAssets)it+ β8Levit+ β9ROAit+ β10CFOit + β11SDRevit+ β12SDCFOit+ εit
Table V: Panel regression model of earnings management on intensity of financial crisis
To conclude, the above presented findings do not provide conclusive evidence on whether earnings management activities increase as the intensity of the European Sovereign Debt crisis increases, or decrease otherwise. However, the analysis revealed conclusive evidence that the economic sentiment in the economy has an impact on earnings management activities, even if this impact is not in line with our expectations.
23 V. Conclusion
On the background of the recent economic crisis, the focus of this study is on the impact of different levels of intensity of the crisis on earnings management activities. The expectations this study forms around the topic is that earnings management activities increase as the financial turmoil increases in intensity, and decrease otherwise. This argument is backed up by several ideas as follows.
Firstly, a loss of confidence in the market is suggested to drive up earnings management activities due to the inability of investors to rely on the “gatekeepers” in the economy - that is on auditors, financial intermediaries and others – and on the market itself. Secondly, the world of accounting practice is also suggested to have a contribution in the level of earnings management activities in the economy. This view is supported by both the space within accounting principles that managers have to mend the timely recognition of earnings, and the inability of accounting professionals to enforce rigorous corporate disclosure of company data. Lastly, manager’s personal and financial objectives, together with a period of financial distress might be factors that drive them to engage in earnings management activities in order to ensure the survival of the company.
In order to determine whether or not there is conclusive evidence to support the hypothesis of this study, yearly company data of small, mid and large capitalization firms was analyzed for a period of 6 years. The findings of this study did not provide sufficient evidence to conclude that earnings management activities increase as the intensity of the European Debt Crisis increase, or they decrease as the intensity of the crisis decreases.
However, the study has certain limitations that could represent a starting point for future research. First, the sample of data is not very homogenous in terms of the participation of each country. By restricting the sample size to isolate the effects of the European Debt Crisis on only countries severely affected by it, more conclusive results might be obtained. Lastly, the analysis might provide more interesting insights by using a different proxy for earnings management activities.
24 Bibliography
Arnold, P. J. (2009). Global financial crisis: The challenge to accounting research . Accounting,
Organization and Society 34, pag. 803-809.
Bull, R. (2009). The global financial crisis and the efficient market hypothesis: what have we learned? . Journal of Applied Corporate Finance, 1-26.
Choi, J.-H., Kim, J.-B., & Lee, J. J. (2011). Value relevance of discretionary accruals in the Asian financial crisis of 1997-1998. Journal of Accounting and Public Policy, 166-187. Coffee, J. C. (2009). What went wrong? An initial inquiry into the causes of the 2008 financial
crisis. . Journal of Corporate Law Studies , 1-22.
Commission, E. (2015, February 26). Economic and Financial Affairs. Retrieved from European Commission: http://ec.europa.eu/economy_finance/db_indicators/surveys/index_en.htm Dechow, P. M. (1994). Accounting earnings and cash flows as measures of firm performance:
The role of accounting accruals. Journal of Accounting and Economics 18, pag. 3-42. Dechow, P. M., & Dichev, I. D. (2002). The quality of accruals and earnings: the role of accrual
estimation errors. The Accounting Review, 35 - 59.
Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting Earnings Management. The
Accounting Review, 70(2), 193-225.
Degeorge, F., Patel, J., & Zeckhauser, R. (1999). Earnings management to exceed thresholds.
The Journal of Business , 1-33.
European Sovereign Debt Crisis. (2015). Retrieved from Investopedia:
http://www.investopedia.com/terms/e/european-sovereign-debt-crisis.asp Eurostat. (2015, February). Eurostat. Retrieved from European Commission:
http://ec.europa.eu/eurostat/tgm/table.do?tab=table&plugin=0&language=en&pcode=teib s010
Filip, A., & Raffournier, B. (2014). Financial Crisis And Earnings Management: The European Evidence. The International Journal of Accounting, 49(4), 455-478.
Gaio, C. (2010). The relative importance of firm and country characteristics for earnings quality around the world. European Country Review, 19(4), 693-738.
Gorgan, C., Gorgan, V., Dumitru, V. F., & Pitulice, I. C. (2012). The evolution of the accounting practices during the recent economic crisis: Empirical survey regarding the earnings management. Amfiteatru economic, XIV, 550-562.
25
Hakkio, C. S., & Keeton, W. R. (2009). Financial Stress: What Is It, How Can It Be Measured, and Why Does It Matter? Economic Review, 94(2), 5-50.
Healy, P. M., & Wahlen, J. M. (1999, December ). A review of the earnings management
literature and its implications for standard setting. Accounting Horizons, vol. 13, nr.4, pp. pag. 365-383.
Hollo, D., Kremer, M., & Lo Duca, M. (2012). CISS - A composite indicator of systemic stress in the financial system. Working Paper Series , 1426, 1-49.
IAS 18 - Revenue. (2014). Retrieved from Delloite - IAS Plus:
http://www.iasplus.com/en/standards/ias/ias18
Iatridis, G., & Dimitras, A. I. (2013). Financial crisis and accounting quality: Evidence from five European countries. Advances in Accounting, incorporating Advances in International
Accounting, 29, 154-160.
Jenkins, D. S., Kane, G. D., & Velury, U. (2009). Earnings Conservatism and Value Relevance across the business cycle. Journal of Business Finance and Accounting, 36(9) & 10, 1041-1058.
Johnson, M. F. (1999). Business cycles and the relation between security returns and earnings .
Review of Accounting Studies, 4, 93-117.
Jones, J. J. (1991). Earnings management during import relief investigations. Journal of
Accounting Research, 29(2), 193-228.
Kousenidisa, D. V., Ladas, A. C., & Negakis, C. I. (2013). The effects of the European debt crisis on earnings quality. International Review of Financial Analysis, 30, 351-362. Lane, P. R. (2012). The European Sovereign Debt Crisis. Journal of Economic perspectives,
26(3), 49-68.
Limited, S. (2015). STOXX Indices. Retrieved from STOXX:
http://www.stoxx.com/indices/index_information.html?symbol=SXXNFE Magnan, M. L. (2009). Fair Value Accounting and the Financial Crisis: Messenger or
Contributor? Accounting Perspectives, 189-213.
Sloan, R. G. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings? . The Accounting Review, 289 - 315.
Stolowy, H., Lebas, M. J., & Ding, Y. (2013). Financial Accounting and Reporting - A global
26
Subramanyam, K. R. (1996). The pricing of discretionary accruals. Journal of Accounting and
Economics, 249 - 281.
Trombetta, M., & Imperatore, C. (2014). The dynamic of financial crises and its non-monotonic effects on earnings quality. Journal of Accounting and Public Policy, 33, 205-232. Watt, A. (2008). The economic and financial crisis in Europe: addressing the causes and the
repercussions. Munich Personal RePEc Archive 3, pag. 1-23.
Zang, A. Y. (2012). Evidence on the Trade-Off between Real Activities Manipulation and Accrual-Based Earnings Management. The Accounting Review, 675-703.
