MSc Accountancy & Control, variant Accountancy Faculty of Economics and Business, University of Amsterdam
Master Thesis
Changes in SFAS 130: Investors’ reaction on U.S. firms’ Comprehensive
Income reporting location: An extension of Bamber et al. (2010).
Name: Wendy Rossenaar
Student number: 6043550
Date: June 17, 2014
First supervisor: Dr. A. Sikalidis
Second supervisor: Dr. B. Qin
Table of Contents
1 Introduction……….4
2 Literature Review……….7
2.1 Comprehensive Income………..………..7
2.2 Updated SFAS 130 (2011-5)……….………...8
2.3 Efficient Market Theory………...………9
2.4 Positive Accounting Theory………..……….10
2.5 Hypotheses……….…….11
3 Research Method………....15
3.1 Sample Selection…..………...15
3.2 Methodology………...…16
4 Results for testing H1a and H1b……...………21
4.1 Preliminary tests……….21
4.2 Descriptive Statistics……….……….22
4.3 Regression Results………..24
5 Results for testing H2a and H2b…………....………...28
5.1 Preliminary tests……….28
5.2 Descriptive Statistics………..28
5.3 Regression Results………..29
6 Results for testing H3a and H3b………...………33
6.1 Preliminary tests……….33 6.2 Descriptive Statistics………..33 6.3 Regression Results………..35 7 Additional analysis………..………...38 8 Conclusion………...…42 Appendix: References………...………...……….………45
Abstract
This research focuses on the investors’ reaction on the reporting location of other comprehensive income. In 2011 SFAS 130 changed which means that firms are required to report other comprehensive income in a statement of comprehensive income, which is shown in the profit and loss statement. Before 2012 firms had two options, namely reporting other comprehensive income in a statement of shareholders’ equity on the balance sheet or reporting it in the profit and loss statement. According to Bamber et al. (2010) before the change of SFAS 130 managers reported other comprehensive income mostly on the balance sheet since they believed this is less visible to investors. They assume that the higher perceived volatility of other comprehensive income could decrease the performance of the firm when this would be more visible to investors. This research focuses on the actual reaction of investors on the reporting location of other comprehensive income in 2000 (since SFAS 130 was introduced in 1998 and this makes the results comparable to the results of Bamber et al. (2010)) and compares this with the investors’ reaction after the change of SFAS 130 in 2012. The results show that investors do not react differently on the reporting location before the change since stock returns do not differ significantly between the two reporting location choices in 2000. When it became required to report other comprehensive income in a statement of comprehensive income in the profit and loss statement stock returns decreased significantly which implies that managers assumptions regarding reporting other comprehensive income proved to be true: Reporting other comprehensive income in a statement of profit and loss increases transparency to investors which reduces stock returns.
This research contributes to the existing literature on the reporting location of other comprehensive income since it shows empirical evidence on investors’ reaction on the introduction of the change in SFAS 130. Furthermore this study adds to the research of Bamber et al. (2010) since they studied the considerations of managers when reporting other comprehensive income and this study focuses on the ex-ante validity of these considerations in 2000. The ex-ante validity of managers’ considerations is also studied for the period after the change in 2012.
Key words: Other comprehensive income, comprehensive income, reporting location, SFAS 130
1. Introduction
Bamber et al. (2010) investigated how the reporting location of comprehensive income is affected by job security and equity incentives of managers in a sample of U.S. firms. Before 2012, according to SFAS 130, the FASB encouraged enterprises to report comprehensive income in a performance statement (profit and loss), but allowed the alternative of reporting comprehensive income in a statement of equity (Lee et al., 2006, p. 658). Bamber et al. (2010) looked at the comprehensive income reporting behavior of managers before 2012. They found that managers with stronger equity-based incentives and less job security are significantly less likely to use performance reporting (p. 97). Since managers think the reporting location matters, the aim of this research is to look at the actual reaction of investors on the reporting location to see if the assumption of the managers holds. Furthermore, in 2011 SFAS 130 of U.S. General Accepted Accounting Principles (hereafter, U.S. GAAP) has changed. According to the new regulation it is not allowed anymore to report comprehensive income in the statement of shareholders’ equity (Financial Accounting Standards Board , hereafter FASB, 2011, p. 7). This is also incorporated in this research by looking at the investors’ reaction in this situation, namely in 2012. So the objective of this study is to provide empirical evidence regarding the investors’ reaction on the reporting location of comprehensive income of U.S. firms before the change of SFAS 130 (2000) and after the change, that became effective in 2012.
According to FASB, other comprehensive income includes “the change in equity of a business enterprise during a period from transactions and other events and circumstances from non-owner sources. It includes all changes in equity except those resulting from investments by owners and distributions to owners” (FASB 1985, paragraph. 70). Examples of components of other comprehensive income are gains/losses on pensions, gains/losses on available-for-sale securities and foreign currency translations. The question regarding the reporting location of comprehensive income arose from the introduction of SFAS 130. Before 2012, there were two choices regarding the reporting location of other comprehensive income, namely in a statement of equity or in a performance statement (Lee et al, 2006, p. 658). Since the change of SFAS 130 this is not allowed anymore and all items of comprehensive income should be reported in a statement in the profit and loss statement.
studies of Bamber et al. (2010) and Lee et al. (2006). What came from the paper of Bamber et al. (2010) is that managers do act as if they believe that comprehensive income reporting location matters (p. 97). Lee et al. (2006) focused on the reporting decisions of property-liability insurers and found that insurers with a reputation for poor disclosure quality and insurers with a tendency to manage earnings through comprehensive income are more likely to report comprehensive income in a statement of equity. This research showed that when managers think comprehensive income can have a negative impact on the rewards or reputation of the managers, they are more likely to report comprehensive income in the statement of equity. For them it is more harmful to report more transparently (via the performance statement) than to report it in equity. Both studies show that managers choose the reporting location based on personal incentives: these managers assume that when comprehensive income is reported in the performance statement, this is more transparent to the investors and this can influence their job security or personal benefits in a negative way. This would be the case if other comprehensive income shows negative results, since this would be easier detected by investors.
This study looks at the actual reaction of investors on the reporting location of other comprehensive income. The theoretical background involves two theories: the efficient market theory and the positive accounting theory. While the efficient market theory assumes that reporting location should not matter since all information is stated in the financial statements, the positive accounting theory predicts the actions of firm managers as a response to a proposed new accounting standard (Scott, 2012, p. 304). From the perspective of the positive accounting theory, investors should react to a different reporting location, because managers do care about this reporting location and have certain reporting incentives. Both theories are considered during this research.
The sample of this research consists of firms from the Standards & Poor’s (S&P) 500 Index and consists of 262 and 311 firms, representing years 2000 and 2012 respectively. During this research no support is found for the hypothesis that the reporting location matters in 2000, in which there was no required reporting location. Actually, after the change of SFAS 130 in 2011 stock returns have declined significantly which shows that the reporting location of other comprehensive income matters.
The contribution of this research is twofold: 1) contribution to the existing literature in accounting research and 2) societal contribution. In particular, the contribution to the existing
accounting literature is in three ways. First, this research contributes to the literature concerning comprehensive income since it provides empirical evidence regarding comprehensive income reporting location. This is especially the case, since there have not been many quantitative researches regarding this topic, except for the previous mentioned researches of Bamber et al. (2010) and Lee et al. (2006). Second, there has not been an extension of the paper of Bamber et al. (2010) conducted in the same period as the research sample of Bamber et al. (2010), namely in 2000. This research extends the research of Bamber et al. (2010) since it looks at the actual reaction of investors on the reporting location of other comprehensive income. Such a research has not been conducted before. Third, there is no empirical evidence regarding the effect of comprehensive income reporting location after 2011. In 2011 SFAS 130 of U.S. GAAP was announced, in which the choice to report other comprehensive income in a statement of shareholders’ equity or in a performance statement was removed. Instead of this reporting choice firms are required to report other comprehensive income in a statement of comprehensive income (performance statement). By this research the effect of the implementation of this rule can be measured, which has not been done in previous researches.
The societal contribution of this research is about the behavior of managers and FASB. According to Bamber et al. (2010) managers assume reporting location matters, so this research examines the validity of managers' ex-ante reporting choice. Since FASB announced the regulation regarding the reporting location in 2011, the ex-ante validity of the assumptions of managers can be tested in two ways. First, the reaction of investors on the reporting location in the period of 2000 is assessed, which is an extension of the paper of Bamber et al. (2010). Second, the consequences of the new regulation are examined by looking at the volatility of the income shown in a performance statement and the stock returns of public firms after 2011.
The next section includes the literature review, where the concept and the rules regarding comprehensive income are discussed. In this section, a brief analysis of the efficient market theory and the positive accounting theory is also presented. This section ends with the development of the hypotheses. The third section describes the research method, namely the sample selection process, variables and empirical models used. The fourth, fifth and sixth section provides the descriptive statistics and results of the research. The seventh section shows the results of the additional analyses, while the paper ends with a conclusion and recommendations
2. Literature Review 2.1 Comprehensive Income
According to the FASB, U.S. GAAP follow the all-inclusive concept of reporting, which means that all items of income and expenses flow through the income statement. Till the end of 2011 there was an exception for items that reflect changes in assets or liabilities that would not be reported in the income statement but in a separate statement of equity. These items are classified as ‘other comprehensive income’ (Carmichael et al, 2007, p. 353). According to the FASB (2011, p. 17) other comprehensive income consists of the following components:
a) Foreign currency translation adjustments
b) Gains and losses on foreign currency transactions
c) Gains and losses on intro-entity foreign currency transactions that are of long-term investments nature.
d) Gains and losses (effective portion) on derivative instruments that are designated as, and qualify as, cash flow hedges
e) Unrealized holding gains and losses on available –for-sale securities
f) Unrealized holding gains and losses that result from a debt security being transferred into the available-for-sale category from the held-to-maturity category
g) Amounts recognized in other comprehensive income for debt securities classified as available-for-sale and held-to-maturity related to an other-than-temporary impairment recognized
h) Subsequent decreases (if not an other-than-temporary impairment) or increases in the fair value of available-for-sale securities previously written down as impaired
i) Gains or losses associated with pension or other postretirement benefits
j) Prior service costs or credits associated with pension or other postretirement benefits k) Transition assets or obligations associated with pension or other postretirement benefits The importance of reporting other comprehensive income can be determined by looking at the value relevance and decision usefulness of other comprehensive income. Value relevance shows to what extent financial statement information assists investors to predict future firm value by looking at the change in for example earnings or share prices (Scott, 2012, p. 208). In the research of Cahan et al. (2000) the value relevance of reporting comprehensive income is determined. The results show that when comprehensive income is reported in the equity
statement, this does not add any value compared to the income statement (the intended contribution of SFAS 130). On the other hand Kanagaretnam et al. (2009) showed that net income is a better predictor of future net income, which indicates that net income has a higher decision usefulness. This means that reporting other comprehensive income in a performance statement enhances the usefulness of financial statements, compared to reporting it in an equity statement.
2.2 Updated SFAS 130 (2011-5)
In 2011 the FASB issued an Accounting Standards Update regarding comprehensive income (topic 220). This was done to improve the comparability, consistency and transparency of financial reporting and to increase the prominence of items reported in other comprehensive income. Furthermore the FASB considered this measure as a way to facilitate convergence of U.S. GAAP and IFRS (International Financial Reporting Standards). The FASB (2011) decided that all changes in other comprehensive income are not allowed to be reported as part of the statement of changes in stockholders’ equity. Now an entity has the option to present the total of comprehensive income (net income and other comprehensive income) in two ways. For both ways the entity is required to present each component of net income along with total net income, each component of other comprehensive income along with a total for other comprehensive income and a total for comprehensive income. For the style of presentation there are two choices, namely:
1. A single continuous statement in which all above mentioned components should be reported separately.
2. A two-statement approach in which the entity is required to present the components of net income and total net income in one statement. Furthermore the statement of other comprehensive income should immediately follow from the statement of net income in which also a statement of comprehensive income is presented.
Entities are still required to make reclassification adjustments when items are reclassified from other comprehensive income to net income. This requirement does not differ by reporting choice. Also, the update of SFAS 130 does not change the items that must be reported in other comprehensive income, as mentioned in paragraph 2.1. Furthermore the amendments should be
period) after December 15 2011, where they should be applied for private entities for fiscal years ending after December 15 2012.
The update results from a project of the FASB and the IASB (International Accounting Standards Board) to look for convergence of U.S. GAAP and IFRS. Under IFRS the same rules regarding the choice of presenting comprehensive income in a single continuous statement or a two-statement approach hold. The amendment results in more converged guidance on how other comprehensive income is presented under both standards. Actually there are still differences in reporting other comprehensive income, that are not diminished because of the change in presentation format.
2.3 Efficient Market Theory
The efficient market theory is an important normative theory that can be applicable to the question posed in this research. This theory mainly means that security prices fully reflect all available information. This is the case since prices reflect information to the point when marginal benefits of acting on information do not exceed the marginal costs (Fama, 1991). According to this theory investors should not react differently to the presentation of comprehensive income in the two forms, since all information is available and visible to the investors. The efficient market theory implies that it is the information content of disclosure, not the form of disclosure itself, that is valued by the market (Scott, 2012, p.108). Regarding this theory, different forms can be distinguished that are described by Scott (2012). The strong form claims that prices instantly reflect all available information about a firm, even insider information. The semi-strong form claims that prices reflect all publicly available information (Scott, 2012, p.110). Hereafter when referred to market efficiency, it is about semi-strong efficiency in which insider information is not available for the investors.
There has been a research before in which the efficient market theory was applicable to the topic of this research. Dehning and Ratliff (2004) did this research about the effectiveness of SFAS 130. They looked at the possible effect of the accounting change from SFAS 115 to SFAS 130, in which it became required to report a statement of comprehensive income in the financial statements. They found, since markets should be efficient, that when the disclosure rules of comprehensive income change this does not affect firm value. According to them the market does
not value information differently solely because of a reporting location, which is in line with the efficient market theory.
2.4 Positive Accounting Theory
The positive accounting theory is another theory that is applicable to this field of study. Contrary to the efficient market theory, the positive accounting theory is a positive theory; a theory that states what the manager will do, given the applicable laws. The difference between the normative theory (efficient market theory) and a positive theory (positive accounting theory) is one of the most important differences on which the hypotheses of this study are based.
The definition of positive accounting theory, described by Scott, is the following: ‘It is concerned with predicting such actions as the choices of accounting policies by firm managers and how managers will respond to proposed new accounting standards’ (Scott, 2012, p.304). This is mostly based on contracts. These contracts contain different (mostly) financial accounting variables which influence the behavior of managers. There are three types of contracts that can influence the behavior regarding accounting choice, namely contracts based on the bonus plans hypothesis, debt covenant hypothesis and political cost hypothesis. These types of contracts are interpreted by managers in the most optimal way: The most optimal way for the firm (efficiency) or the most optimal way for themselves (opportunistic). In case of other comprehensive income reporting location, the managers will mostly base their decision on the way it can be most optimal for themselves (which can be optimal for the firm in an indirect way). In the case of other comprehensive income reporting location the three hypotheses can be interpreted in the following way:
· Bonus plans hypothesis: This hypothesis states that firms with bonus plans are more likely to shift income from future to current period, since managers would like to have high remuneration. Clearly, this hypothesis is applicable to the case of the managers described in the paper of Bamber et al. (2010), since these managers try to increase their remuneration by showing other comprehensive income at a specific reporting location. · Debt covenant hypothesis: This hypothesis states that firms closer to violation of debt
covenants are more likely to shift income from future to current period. This hypothesis is also applicable to the managers (but to a lower extent than the bonus plan hypothesis),
since debt contract violation can also be a reason to report comprehensive income at a specific reporting location.
· Political cost hypothesis: This hypothesis states that the greater political costs are, the more likely a firm will defer current income to future periods. This seems less applicable to the managers described by Bamber et al. (2010), since in this research these managers are not asked about this type of reasons. This hypothesis is more based on reasons benefiting for the firm, instead of a reason for managers to profit themselves.
Comparing these three hypotheses, the bonus plan hypothesis is most applicable to this research question. Namely, managers try to optimize their remuneration and (the volatility of) comprehensive income can have a great influence on that, since according to Bamber et al. (2010) comprehensive income is seen as more volatile than net income. The research of Graham et al. (2005) shows that as well: Managers think the volatility of comprehensive income can damage the stock price of the firm since a higher volatility of perceived as a higher risk. When other comprehensive income is reported in the performance statement (which is more visible for investors), this can have a negative impact on the performance and remuneration of managers.
According to these researches managers try to optimize their performance (because of bonus incentives) and they are afraid of other comprehensive income decreasing their performance measurement: Investors should take this behavior into account. This can result in a positive or negative reaction, measured by a change in stock returns (which is discussed in the next part).
2.5 Hypotheses
According to previous literature there have been several studies regarding the importance of comprehensive income in general, importance for investors and the function and usefulness of the reporting location. Since there are many different outcomes of these researches, this makes it hard to predict the outcome of this research. Regarding the hypotheses, it is important to consider one aspect of reporting comprehensive income: the volatility of the firms’ performance. Since the FASB wrote an Exposure Draft regarding obligatory reporting location of other comprehensive income in a single continuous statement, most comments were about the perceived increasing volatility of firms’ performance. Managers assume a higher perceived volatility of the firms’
performance (when other comprehensive income is shown in a performance statement) can decrease the stock prices of the firm since investors will react to that (Graham et al., 2005). This theory is also supported by other behavioral finance researches and laboratory experiments (Bamber et al., 2010). To develop the hypotheses, two theories are used, namely the efficient market theory and the positive accounting theory which are discussed before. When these theories are applied to this research, they point to different possible outcomes.
The efficient market theory would predict that investors would not react differently to the reporting location, since all information is visible. This is the normative approach, which does not predict what the actual reaction of the investors will be, according to the given information, but what the reaction should be. According to this theory, stock returns and the volatility of the income shown in the performance statement would be equal under each reporting location since investors do not value the reporting location. This theory would also imply that the stock prices and the volatility of the income shown in the performance statement would not be different in the period after 2011.
On the contrary, the positive accounting theory would predict that investors would react differently. Since managers behave in the most optimal way to enhance their revenues by showing comprehensive income in the equity statement to ‘hide’ things that could lower their revenues, since they assume this is less visible in the statement of shareholders’ equity. According to Hirst and Hopkins (1998) and psychological researches, the analysis of investors is effected by the clarity and transparency of the disclosure, which will be diminished by reporting other comprehensive income in the equity statement. This would lead to a situation when other comprehensive income is reported in the equity statement (where it is less visible and the meaning of other comprehensive income will be understood to a lesser content); this would result in higher stock returns. This is in line with the expectations of the managers.
Comparing these theories, it is useful to look which theory is mostly applicable to this situation. Since the efficient market theory does not hold for the behavior of managers, it seems likely that investors will respond to the behavior of managers as well according to the positive accounting theory. So the positive accounting theory, which is a behavioral theory, is more applicable to this topic. Unless the efficient market theory could hold for a rational and objective investor, it seems more likely that this is not the case. This is also likely since the goal of the
change in SFAS 130 is to increase transparency. Applying the positive accounting theory on this question results in the following hypotheses:
The first hypotheses (1a and 1b) are about the volatility of the income reported in the performance statement and the variability in stock returns in 2000. According to the positive accounting theory, the reporting location would matter for investors since it matters for managers. This would imply that the volatility of the income reported in the performance statement would be higher when other comprehensive income is reported in the performance statement. Regarding the second part of the hypothesis this would imply that stock returns would be lower when other comprehensive income is reported in the statement profit and loss. This results in the following hypotheses:
H1a: In 2000 volatility of the income reported in the performance statement is higher when comprehensive income is reported in a performance statement.
H1b: In 2000 stock returns are lower when other comprehensive income is reported in a performance statement.
Besides doing the extension of Bamber et al. (2010) other tests are performed to test if the reporting location matters. When managers had the possibility to report other comprehensive income in a statement of shareholders’ equity on the balance sheet or in the profit and loss statement, most of the firms reported other comprehensive income in a statement of shareholders’ equity. To create a bigger sample, the second pair of hypotheses tests if the reporting location matters, which is done by including the sample of 2012. This enhances the suitability of the research. The second pair of hypotheses looks like the following:
H2a: The volatility of the income reported in the performance statement is higher when comprehensive income is reported in a performance statement.
H2b: Stock returns are lower when other comprehensive income is reported in a performance statement.
Since U.S. GAAP changed in 2011, the influence of reporting location changed as well. During this research the impact of a required reporting location of comprehensive income is measured. A comparison between the two periods (2000 and 2012) is done. According to the assumption of managers that reporting other comprehensive income in the performance statement increases the volatility of the income presented in the performance statement or reduces stock returns, a required reporting location (in a single continuous statement) would result in a higher volatility or lower stock return as well. This is in line with the positive accounting theory. The previous mentioned assumptions results in the following hypotheses:
H3a: Volatility of the income reported in the performance statement is higher for the period after implementing the required reporting location.
H3b: Stock returns are lower for the period after implementing the required reporting location.
When it comes out that investors do not react differently to the reporting location, it will seem more likely that the efficient market theory holds.
3. Research Method
3.1 Sample selection
The sample consists of firms from the S&P 500. The S&P 500 is one of the most commonly used benchmarks for the overall U.S. stock market. This index is made to reflect the risk/return characteristics of the large cap universe. Using firms listed in the S&P 500 makes it possible to compare the results of Bamber et al. (2010) to the regression analysis used to test for the first hypothesis (hypothesis 1a and 1b). Furthermore this sample is really useful to test the third (3a and 3b) hypothesis, since the change in regulation is about U.S. GAAP and the sample consists of U.S. firms.
Data is collected from January 2000 on, for a period of one year (fiscal year of 2000), since SFAS 130 is introduced in the end of 1998. This was also the research period Bamber et al. (2010) used for their research which makes these researches comparable. For the regression analysis regarding the second hypothesis the period after December 15 2011 is used, since from that day on firms are required to report comprehensive income in a single continuous statement. For this analysis a period of one year is used too (fiscal year of 2012). Annual reports of 2013 were not available during the research period.
The sample composition that is used to test the first hypotheses (1a and 1b) is shown in table 1. First the firms of which no annual report could be found in Company.info were dropped from the sample. This was mainly due to takeovers and firms that went bankrupt. Second firms that do not report other comprehensive income or do not specify other comprehensive income were dropped from the sample. Third there were some firms dropped since there was no information available about historical stock prices. Finally since all firms in the S&P 500 have to report under U.S. GAAP, no firms had to be removed for reporting under another standard.
Table 2 shows the sample composition of firms in 2012. This sample, together with the sample shown in table 1 is used to test for hypotheses 2a, 2b, 3a and 3b. Similar to the sample of 2000, first the firms of which no annual reports were available in Company.info were removed from the sample. Second there were some firms of which information in Compustat was missing; these firms were removed as well. Third there were firms removed that had a fiscal year ending before December 15, which made them not useful for this research since they were not required to report under the renewed SFAS 130. Fourth there were again firms that did not report other
comprehensive income or did not specify it. Finally there were some firms of which other data was missing. This results in the following samples:
Table 1. Sample selection (2000)
All shares listed on S&P 500 500
- Firms of which no annual reports was available - 161
- Firms that do not report OCI - 44
- Firms with no specification of OCI - 26
- Firms with no information on stock prices - 7 Sample size for regressions to test for H1a and H1b 262
Table 2. Sample selection (2012)
All shares listed on S&P 500 500
- Firms of which no annual report was available - 112 - Firms of which information is missing in Compustat - 33 - Firms with a fiscal year ending before 15-12 - 29
- Firms with no information on OCI - 8
- Firms with no OCI - 4
- Firms of which data is missing - 3
Sample size for regressions to test for H2a, H2b, H3a and H3b 311
3.2 Methodology
To test the hypotheses several regression models are used. To test the first hypotheses (1a and 1b) a model is used to test for the actual reaction of investors in the period 2000, measured by changes in the volatility of the income reported in the performance statement and changes in stock returns. To test for the second pair of hypotheses another analysis is done to look at the differences in reporting location, in which the sample of 2012 is included as well. To test these hypotheses the same model is used as the one used to test the first hypotheses. After these regression analyses a third analysis is done (with a different model) to look at changes in volatility of the income presented in the performance statement and changes in stock returns in the period of 2012 compared to the period of 2000.
income influences the volatility of the income reported in the performance statement. In case a firm reports other comprehensive income on the balance sheet, the performance statement only shows ‘net income’. When other comprehensive income is shown in the ‘profit and loss statement’, the performance statement shows ‘net income’ plus ‘other comprehensive income’. The difference in volatility of these two financial statements items is used to test if there is a difference in the volatility because of the reporting location. Bamber et al. (2010) also use the volatility of comprehensive income as one of the control variables because the managers assume a higher volatility of comprehensive income causing a higher perceived risk (by investors) resulting in lower stock returns. For that reason stock return is used as another dependent variable, since managers think that the reporting location of other comprehensive income influences stock returns.
The regression model tests if these assumptions from managers are true according to the volatility of the income presented in the performance statement and the variability of stock returns (hypotheses 1a and 1b). The dependent variables of this model are the reporting location choice and other control variables. The choice for the control variables is made mostly based on previous researches of Bamber et al. (2010), Lee et al. (2006) and Turktas et al. (2013). The researches of Bamber et al. (2010) and Lee et al. (2006) tested the reporting choice of managers while the research of Turktas et al. (2013) also tested the validity of the comprehensive income location choice of managers in Europe. Combining these researches results in the following model of control variables that influence the volatility of the income presented in the performance statement and the variability in stock returns.
Vol.PS or STOCK.RETURNS = β0 + β1 LOCATION + β2 OCI + β3 NI + β4 Vol.OCI + β5
MKTSEC + β6 PENSION + β7 FORCUR + β8LEVERAGE + β9LSIZE + ε.
Equation (1)
The first independent variable is the reporting location, which is the most important variable for this regression. A dummy-variable equal to 1 when comprehensive income is reported in a single statement of comprehensive income (profit and loss) is used for this, 0 when other comprehensive income is reported in a statement of shareholders’ equity. For the control
variables, the research of Turktas et al. (2013) is most comparable, so most variables are also used in the research of Turktas et al. (2013).
First the scaled amount of other comprehensive income is added since the results of Lee et al. (2006) showed that the relative amount of other comprehensive income is value relevant for investors. Furthermore Goncharov and Hodgson (2011) show that net income is a decision relevant metric for investors so the relative size (which is done by scaling net income to assets) is added as a control variable. The volatility of other comprehensive income is also used as a control variable. Since this relationship is important for the managers in their decision-making regarding the reporting location (Bamber et al., 2010) this variable is also added to the research.
Subsequently the different components of other comprehensive income are added as control variables. Bamber et al. (2010) added these variables in their research since according to an analysis of the comment letters on FAS No. 130 almost twice as many critics express concern about the effect of unrealized gains and losses on AFS (Available-for-sale) than about the other components. This can be taken into consideration by managers when considering the reporting location which can influence investors’ reaction. Also leverage is added as a control variable, since Graham et al. (2005) found that managers of more levered firms are more concerned with smoothing earnings to minimize the perceived risk of a firm. In that case the effect of this variable would be that managers of a more levered firm would be more concerned about the volatility of comprehensive income. Furthermore like Lee et al. (2006) and Bamber et al. (2010) did, a control variable for the size of the firm is added. In Table 3 all variables are shown with a short description.
During the research the variable ‘Vol.PS’ is also used as an independent variable since managers assume that the higher perceived volatility of the income shown in the performance statement can results in lower stock returns. The additional equation used to test for hypothesis 1b and 2b changed into the following equation:
STOCK.RETURNS = β0 + β1 Vol.PS + β2 LOCATION + β3 OCI + β4 NI + β5 Vol.OCI + β6
MKTSEC + β7 PENSION + β8 FORCUR + β9LEVERAGE +β10LSIZE + ε.
To test the third pair of hypotheses (H3a and H3b) almost the same variables are used as when the first hypotheses (H1a, H1b, H2a and H2b) were tested. The dependent variables are again the volatility of the income presented in the performance statement and the variability of stock returns. Regarding the control variables only the variable ‘Location’ is removed, since the reporting location is fixed from 2012 on and a dummy variable is added for this. This dummy variable will take the value of 1 when the reporting year is 2012, 0 when the reporting year is 2000. By using this option, it is possible to see what the consequence is of a required reporting location. This results in the following regression model:
VOL.PS or STOCK.RETURNS = β0 + β1 YEAR + β2 OCI + β3 NI + β4 Vol.OCI + β5
MKTSEC + β6 PENSION + β7 FORCUR + β8LEVERAGE + β9LSIZE + ε.
Equation (3)
Furthermore, equation 3 also changed when Vol.PS was added as an independent variable:
STOCK.RETURNS = β0 + β1 Vol.PS + β2 YEAR + β3 OCI + β4 NI + β5 Vol.OCI + β6
MKTSEC + β7 PENSION + β8 FORCUR + β9LEVERAGE +β10LSIZE + ε.
Equation (4)
Currently several researches report on the effect of the financial crisis. It has been considered in this research to add a variable for the period of financial crisis, but according to prior research the two research periods of this research did not take place in a situation of financial crisis. Actually there is no consensus about the exact time frame of the financial crisis, but most researches do agree on the assumption that the crisis started in 2007 and ended in 2009. (Francis, 2013 and Fahlenbrach & Stulz, 2011).
To test the suitability of the different models, before the descriptive statistics and the regression results are shown, the results of other different tests are shown. These tests test for the Gauss-Markov assumptions, linearity and multicollinearity.
Table 3. Explanation of variables Variable Explanation
FORCUR 1 if the foreign currency translation scaled by total assets in the comprehensive income year exceeds the sample median, otherwise 0. Manually collected from annual reports.
LEVERAGE Long-term debt scaled by total assets, as of the end of the comprehensive income year.
LOCATION Way of reporting comprehensive income, with a 1 for firms that report in a single statement and 0 for firms that report in equity. Manually collected from annual reports.
LSIZE Logarithm of the market value of the firms’ common shares outstanding, as of the end of the comprehensive income reporting year.
MKTSEC 1 if the gains or losses of AFS securities scaled by total assets in the comprehensive income year exceeds the sample median, otherwise 0. Manually collected from annual reports.
NI Net income scaled by total assets of the firm for the comprehensive income year.
OCI Other comprehensive income scaled by total assets, calculated as the difference between ‘net income scaled by total assets’ and ‘comprehensive income scaled by total assets’ for the comprehensive income year.
PENSION 1 if the actuarial gains or losses scaled by total assets in the comprehensive income year exceeds the sample median, otherwise 0. Manually collected from annual reports.
STOCKRETURN Stock return of the firm for the comprehensive income year.
Vol.OCI Volatility of other comprehensive income, calculated by dividing the standard deviation of comprehensive income scaled by total assets with the standard deviation of net income scaled by total assets. This is measured over the initial comprehensive income reporting year and the two prior years. Data regarding OCI is manually collected from annual reports.
Vol.PS Volatility of the income presented in the performance statement. In case other comprehensive income is reported at the balance sheet, this includes the volatility of net income. When other comprehensive income is reported in profit and loss, this includes comprehensive income.
YEAR 1 if the comprehensive income reporting year is 2012, 0 if the comprehensive income reporting year is 2000.
4. Results for testing H1a and H1b
4.1 Preliminary tests
To make sure the regression model is verifiable, it has to fulfill the Gauss-Markov assumptions. These assumptions are:
1. The average value of the residuals should be equal to zero (E(ut)=0) 2. The population covariance should be equal to zero (Pop.cov.(ut, xt) = 0) 3. The variances of the sample should be constant (Var(Ut)= σ²)
4. There should not be a situation of autocorrelation (Cov.(ut, uj)=0)
Furthermore the model needs to be tested for linearity which means that there should not be a situation of misspecification (Rossenaar, 2012). Finally a test for multicollinearity is performed.
To test the Gauss-Markov assumptions several tests are performed. To make sure the model fulfills the first assumption a constant factor is added to the model. Regarding the second assumption no test needs to be done as well, since all variables in the models are exogenous which means that the population covariance is equal to zero. To test for the third assumption a White test is performed to test if there is a situation of heteroskedasticity. The result of this test shows that there is no situation of heteroskedasticity: the p-value of the test is 93,87 percent. The last assumption is about time-series which is not applicable to this research since the data is undated. To conclude: the models used in this research fulfill all the Gauss-Markov assumptions.
Two extra tests need to be done as well. First the models are tested for linearity by performing a Ramsey RESET test that used the powers of the independent variables. The p-values of the tests for the model of 2000 is 29,39 percent which shows that there is no situation of misspecification in this model. Finally a test for multicollinearity is done, namely by measuring the variance inflation factor (vif). When this factor is below ten, no situation of multicollinearity exists. The vif-factor of the model of 2000 is 1,13 which shows that there is no situation of multicollinearity.
In the preliminary tests of the other models, only the results of the White-test, Ramsey RESET-test and the Variance Inflation Factor are shown, since the other models have the same characteristics regarding the first, second and fourth Gauss-Markov assumption as this model.
4.2 Descriptive statistics
Table 4 shows the descriptive statistics for the regressions used to test for hypothesis 1a and 1b. As mentioned above, the sample consists of 262 observations. For this regression the dependent variables are the volatility of the income shown in the performance statement and the stock returns. Among other variables, one of the most important independent variables is the reporting location of other comprehensive income. Out of the 262 observations, the biggest part of the firms reported other comprehensive income in a statement of shareholders’ equity, namely 255 (97,33%), while only 7 firms (2,67%) reported other comprehensive income in the statement of profit and loss. The average volatility of the income shown in the performance statement is equal to 0,0242923, while the average stock return in 2000 was equal to 0,0557827. This indicates an average stock return of 5,6 percent. The mean of other comprehensive income scaled by total assets is equal to -0,01219, while the mean of net income scaled by total assets is equal to
Table 4. Descriptive Statistics of the sample used to test for H1a and H1b
Panel A: Sample composition
Method Number of observations Percentages
OCI reported on balance sheet 255 97.33%
OCI reported in profit and loss 7 2.67%
Panel B: Summary statistics
Variable 1st Quartile Median 3rd Quartile Mean Standard Deviation Vol.PS 0.0057286 0.0137289 0.0302613 0.0242923 0.0317584 STOCK RETURN -0.2292971 0.0089109 0.2922645 0.0557827 0.4432932 LOCATION 0 0 0 0.0267176 0.1615653 OCI -0.0124625 -0.0034393 -4.76E-08 -0.0121904 0.0931233 NI 0.0208626 0.0567267 0.1031409 0.0689466 0.0726349 Vol.OCI 0.9433528 1.03813 1.375166 1.796948 3.247952 MKTSEC 0 0 1 0.278626 0.4491811 PENSION 0 0 0 0.110687 0.3143446 FORCUR 0 0.5 1 0.5 0.5009569 LEVERAGE 0.0727506 0.1681834 0.2927784 0.1887844 0.1422814 LSIZE 3.930338 5.032514 5.781755 4.905731 1.214332
0,0689466. This makes the mean of total comprehensive income scaled by total assets equal to 0,0567566, which indicates that net income is responsible for a substantial amount of total comprehensive income. The mean of the volatility of other comprehensive income of this sample is equal to 1,796948 which means that the volatility of comprehensive income is on average nearly 1,8 times as high as the volatility of net income.
Table 5 shows the results of a Spearman correlation test. In this test the variable ‘Volatility of the income shown in the performance statement’ is also included. This is the case, since during the research it showed that the volatility of the income shown in the performance statement influences stock prices, which will come back later in this research in one of the regressions. In the first place there is a positive relationship between the variables ‘NI’ and ‘Vol.OCI’ and ‘Vol.PS’. This means that when NI is higher, this results in a higher volatility of the income shown in the performance statement. In case of OCI, this relation is comparable; When the volatility of OCI is higher, this results in a higher volatility of the income shown in the
Table 5. Correlation of variables used for tests of H1a, H1b, H2a and H2b.
Vol.PS OCI NI Vol.OCI MKTSEC PENSION FORCUR LEVERAGE LSIZE Vol.PS 1 P-value 0 OCI -0.0627 1 P-value 0.3119 0 NI 0.1309** 0.0766 1 P-value 0.0342 0.2163 0 Vol.OCI -0.1119* -0.4607*** -0.0626 1 P-value 0.0705 0.000 0.313 0 MKTSEC -0.0959 0.1108* -0.0848 0.0909 1 P-value 0.1214 0.0733 0.1711 0.1424 0 PENSION -0.0382 0.0106 -0.0983 -0.0505 -0.0293 1 P-value 0.538 0.8647 0.1123 0.4157 0.6367 0 FORCUR -0.069 0.1137* -0.0636 0.0626 0.1788* -0.0608 1 P-value 0.2655 0.0661 0.305 0.313 0.0037 0.3267 0 LEVERAGE -0.1169* 0.0125 -0.122** -0.0625 -0.1973*** 0.1064* -0.1369** 1 P-value 0.0588 0.8405 0.0486 0.3133 0.0013 0.0857 0.0267 0 LSIZE -0.1492** -0.0037 -0.0614 0.0071 0.1149* 0.1143* -0.0736 0.0059 1 P-value 0.0158 0.9529 0.3233 0.9094 0.0638 0.0653 0.2363 0.925 0
performance statement. Furthermore a higher amount of OCI results in a lower volatility of OCI. Next there is a positive relationship between some of the components of other comprehensive income (MKTSEC and FORCUR) and OCI. This means that when these components are bigger, OCI also increases. Furthermore, regarding the components of other comprehensive income, there is also a positive relationship between the size of firms, leverage of firms and the components (MKTSEC and PENSION) of other comprehensive income. This could be because bigger and more levered firms have more resources to invest in pensions and foreign currency transactions, which would result in a positive relationship.
4.3 Regression results
The results of the first regression analysis are shown in Table 6. By running this regression hypothesis 1a, in which the dependent variable is the volatility of the income shown in the performance statement, is tested (equation 1). At first, there is not a significant relationship between the volatility of the income presented in the performance statement and the reporting location of other comprehensive income since the p-value is equal to 0.597. This means that hypothesis 1a should be rejected. Actually, there are other variables that have a significant relationship with the volatility of the income presented in the performance statement. First, the sign of the relationship between OCI (the amount of other comprehensive income scaled by total assets) and Vol.PS is negative. This means that when OCI is bigger, Vol.PS becomes smaller.
Table 6. Regression results for regression 1 (Dependent variable = Vol.PS)
Variable Coefficient Std. Error t P-value
LOCATION -0,0063557 .0120141 -0.53 0,597 OCI -0,0450772 .0238505 -1.89 0,06* NI 0,0416357 .0271128 1.54 0,126 Vol.OCI -0,0016284 .0006785 -2.40 0,017** MKTSEC -0,004202 .0045588 -0.92 0,358 PENSION -0,0011674 .0062077 -0.19 0,851 FORCUR -0,0036174 .0039922 -0.91 0,366 LEVERAGE -0,0291194 .0140508 -2.07 0,039** LSIZE -0,0035049 .0016247 -2.16 0,032** Cons. 0,049785 .0094147 5.29 0,000*** N 262
This result is not in line with the expectations of managers that the volatility of OCI is bigger than the volatility of NI. Second, there is also a negative relationship between Vol.OCI and Vol.PS. This result is comparable to the relationship between OCI and Vol.PS. Third there is a negative relationship between LEVERAGE and Vol.PS. This means that when a firm has a higher relative amount of debt, this results in a lower volatility of the income presented in the performance statement. This is in line with the expectations, since managers of more levered firms are more concerned about the volatility of comprehensive income. Finally there is a negative relationship between LSIZE and Vol.PS. This implies that bigger firms show a lower volatility of the income presented in the performance statement. In short, a big part of the results is not significant among which the most important variable ‘Location’. Some control variables seem to have a significant relationship with Vol.PS.
To test for the second hypothesis (1b) another regression analysis has been performed. In this regression analysis ‘Stock return’ functioned as the dependent variable to measure the effect of the reporting location on the reaction of investors (equation 1). By testing this hypothesis, it shows that there is a less strong relationship between stock returns and the control variables used compared to the previous regression analysis since the p-values became higher. Again the most important dependent variable ‘Location’ is not significant, which means that the reporting location of other comprehensive income does not have an effect on stock returns. Hypothesis 1b
Table 7. Regression results for regression 2 (Dependent variable = STOCK RETURN)
Variable Coefficient Std. Error t P-value
LOCATION -0,0263848 .1702038 -0.16 0,877 OCI 0,1537212 .3378917 0.45 0,650 NI 0,3450063 .3841081 0.90 0,370 Vol.OCI -0,0009598 .0096121 -0.10 0,921 MKTSEC 0,1270127 .064585 1.97 0,050** PENSION -0,0415457 .0879448 -0.47 0,637 FORCUR 0,0918114 .0565578 1.62 0,106 LEVERAGE 0,5909566 .1990585 2.97 0,003*** LSIZE -0,0210622 .0230172 -0.92 0,361 _cons -0,04787 .133379 -0.36 0,720 N 262 R-squared 0,0588
should be rejected. This implies that investors do not react differently between the two reporting location choices. Regarding the other control variables, the regression analysis showed that there are only significant results for the variables ‘MKTSEC’ and ‘LEVERAGE’. Regarding the variable ‘MKTSEC’, the positive coefficient shows that when a firm trades more in available-for-sale securities, this results in higher stock returns. According to Turktas et al. (2013) this can be the case since when a firm invests more in available-for-sale securities, this can mean that this firm has more expertise to make gains on these securities. The other significant variable is LEVERAGE. This shows that a more levered firm shows higher stock returns, which is not the same result as the previous regression analysis. The reason why this can be the case is that managers may be more concerned about the performance of the firm when they have a higher level of leverage. This shows different results than the first regression analysis. It is important to consider that a different dependent variable is used for this regression.
As mentioned earlier in the literature review, managers assume that the higher volatility of other comprehensive income results in lower remunerations for managers. This would be the case, because other comprehensive income would be reported more transparently and this would influence the reaction of investors, in the form of a change in stock prices. In the first test the volatility of the income presented in the performance statement is used as a dependent variable, but this variable can also be used as an independent variable since managers assume that this variable influences the perception of investors when they evaluate the performance of the firm. In that case adding the variable ‘Vol.PS’ could enhance the R2 of the model. In this case the volatility of the income shown in the performance statement could have an influence on stock prices and stock returns. To test if this variable is another useful independent variable a third regression analysis is performed of which the results are shown in table 8.
In this regression analysis ‘Stock return’ functioned again as the dependent variable and equation 2 is used for the regression analysis. The expectation of ‘Vol.PS’ having an influence on stock returns became true; the sign of this variable is negative which means that a higher volatility of the income shown in the performance statement resulted in lower stock returns. Regarding the other variables the results are similar to the regression results of the previous regression analysis. The variable ‘LOCATION’ is again not significant which means that there is
Regarding the other variables this means that there is still a positive relationship between MKTSEC and stock returns and a positive relationship between the amount of debt and stock returns. The R2 of the model has increased, namely from 5.88% in regression 2 to 8.13% in regression 3. This means that adding ‘Vol.PS’ as an independent variable increases the usefulness of the model.
Table 8. Regression results for regression 3 (Dependent variable = STOCK RETURN)
Variable Coef. Std. Err. t P>t
Vol.PS -0,2190394 .8852285 -2.47 0.014** LOCATION -.0403062 .1685871 -0.24 0.811 OCI .0549844 .3368674 0.16 0.870 NI .4362049 .3820296 1.14 0.255 Vol.OCI -.0045268 .0096241 -0.47 0.639 MKTSEC .1178088 .0640439 1.84 0.067* PENSION -.0441027 .0870671 -0.51 0.613 FORCUR .0838879 .0560809 1.50 0.136 LEVERAGE .5271737 .1987367 2.65 0.008*** LSIZE -.0287393 .0229962 -1.25 0.213 Cons. .0611788 .1391992 0.44 0.661 N 262 R-squared 0,0813
5. Results for testing H2a and H2b
5.1 Preliminary tests
The model used to test for hypothesis 2a and 2b has the same characteristics regarding Gauss-Markov assumption 1, 2 and 4. For that reason only the tests for heteroskedasticity, linearity and multicollinearity are needed to be performed. First the White-test is performed: The p-value of the test is equal to 0,0286 which means that there is a situation of heteroskedasticity. For that reason different regression analyses (with robust standard errors), that make an adjustment for heteroskedasticity, are performed for this model. Second the Ramsey RESET test regarding linearity is performed. The p-value of this test is equal to 28.64 percent, which means that there is no situation of misspecification. Finally the test regarding multicollinearity is performed. The vif-value is equal to 1.13 which means that there is no situation of multicollinearity.
5.2 Descriptive statistics
In this section the descriptive statistics and the correlation matrix to test for hypotheses 2a and 2b would be shown. Since the same variables of equation 1 are used for this regression analysis, the correlation matrix shown in section 4.2 also applies to this regression analysis. Table 9 shows the descriptive statistics for the regression analyses done to test for hypotheses 2a and 2b. The sample consists of 573 observations, 262 observations from the year 2000 and 311 observations from the year 2012. For this regression the dependent variables are again the volatility of the income shown in the performance statement and stock returns. The independent variable ‘Location’ is again the most important independent variable. The observations of firms that reported other comprehensive income in a statement of profit and loss in 2000 and the observations of firms that reported other comprehensive income in 2012 get a value of ‘1’ in this situation (318 observations). Firms that reported other comprehensive income in 2000 in a statement of shareholders’ equity get a value of ‘0’ (255 observations). The average volatility of the income shown in the performance statement is equal to 0,020926. This is lower than the volatility of the income shown in the performance statement of 2000 which implies that the volatility of the income shown in the performance statement is lower in 2012. The average stock return is equal to 0,1246 which indicates an average stock return of 12,46 percent. The mean of other comprehensive income scaled by total assets is equal to -0,0063472 while the mean of net
Table 9. Descriptive Statistics of the sample used to test for H2a and H2b
Panel A: Sample composition
Method Number of observations Percentage
2000 - OCI reported on balance sheet 255 44,50%
2000 - OCI reported in profit and loss 7 1,22%
2012 311 54,28%
Panel B: Summary statistics
Variable 1st Quartile Median 3rd Quartile Mean Standard Deviation
Vol.PS 0,0054215 0,0129231 0,0259193 0,020926 0,0260201 STOCK RETURN -0,0869565 0,118021 0,2721355 0,1245662 0,4243378 LOCATION 0 0 1 0,4442509 0,4973157 OCI -0,0068012 -0,0008098 0,0018145 -0,0063472 0,0635298 NI 0,0199043 0,0535827 0,0935323 0,0617568 0,0646124 Vol.OCI 0,9510624 1,056528 1,461831 1,868317 4,443391 MKTSEC 0 0 1 0,2560976 0,436857 PENSION 0 0 1 0,3222997 0,4677647 FORCUR 0 0,5 1 0,5 0,5004361 LEVERAGE 0,0902196 0,1876894 0,2986433 0,2048804 0,1461806 LSIZE 5,003344 5,431693 5,832892 5,261902 0,9648712
income scaled by total assets is equal to 0,0617568. This makes the mean of total comprehensive income scaled by total assets equal to 0,05541. The mean of the volatility of other comprehensive income of this sample is equal to 1,86317. This means that the volatility of comprehensive income is on average nearly 1,9 times as high as the volatility of net income.
5.3 Regression results
The regression results to test for hypothesis 2a are shown in table 10. Similar to the regression first analysis used in section 4, equation 1 is used for this regression analysis. For this regression analysis a test with robust standard errors is used; the significant results are the same as when a regular regression is performed. At first there is not a significant relationship between the reporting location of other comprehensive income and the volatility of the income shown in the performance statement since this relationship shows a p-value of 0,12: hypothesis 2a should be rejected. There is one variable that shows a significant relationship with the volatility of the income shown in the performance statement. This variable is the logarithm of the size of the firm.
There is a negative relationship between the size of the firm and the volatility of the income shown in the performance statement. This means that the volatility decreases when firms become bigger.
Table 10. Regression results for regression 4 with robust standard errors (Dependent variable = Vol.PS)
Variable Coef. Std. Err. t P>t
LOCATION .0035703 .0022927 1.56 0.120 OCI -.0260241 .1047794 -0,25 0.804 NI .006326 .0310529 0.2 0.839 Vol.OCI -.0002976 .0005233 -0,57 0,570 MKTSEC -.0025321 .002585 -.98 0.328 PENSION .0007839 .0020929 0.37 0.708 FORCUR -.0026538 .0022575 -1.18 0.240 LEVERAGE -.0087683 .0093072 -0.94 0.347 LSIZE -.003867 .0016009 -2.42 0.016** Cons. .0432167 .0106731 4.05 0.000*** N 573 R-squared 0,0463
Notes: *** significant at 1 percent level; ** significant at 5 percent level; * significant at 10 percent
To test for hypothesis 2b a new regression analysis has been performed in which ‘Stock Return’ is the dependent variable (equation 1). The results of this regression analysis can be found in table 11. The most important variable (LOCATION) shows a significant negative relationship with ‘Stock Return’. This means that when a firm reports other comprehensive income in a statement of profit and loss, this results in lower stock returns: Hypothesis 2b should be accepted. Other variables that show a significant relationship with stock returns are ‘MKTSEC’ and ‘LEVERAGE’. First there is a positive relationship between the amount of investments in available-for-sale securities and stock returns. This could be explained by the possibility that firms that invest more in available-for-sale securities have a higher expertise, which results in higher stock returns. Second, there is a positive relationship between the amount of leverage of the firm and stock returns. This could be the case, because managers of more levered firms can be more concerned about reporting comprehensive income which can result in higher stock returns.
Table 11. Regression results for regression 5 with robust standard errors (Dependent variable = STOCK RETURN)
Variable Coef.. Std. Err. t P>t
LOCATION -.137692 .0443459 -3.10 0.002*** OCI .2388655 2.159693 0.11 0.912 NI .2047693 .2717042 0.75 0.451 Vol.OCI .0001164 .0014049 0.08 0.934 MKTSEC .0977089 .0429003 2.28 0.023** PENSION -.033231 .0440159 -0.75 0.451 FORCUR .0491015 .0383984 1.28 0.202 LEVERAGE .4669104 .1764244 2.65 0.008*** LSIZE -.0199113 .0226077 -0.88 0.379 Cons. .1449037 .1418513 1.02 0.307 N 573 R-squared 0,0572
Notes: *** significant at 1 percent level; ** significant at 5 percent level; * significant at 10 percent level
Similar to the regression analyses used to test for the first hypotheses (H1a and H1b) a third regression analysis is performed to include the effect of the volatility of the income shown in the performance statement on stock returns (equation 2). The results of this regression analysis can be found in table 12. Comparing the R-squared of the two regression analyses shows that the R-squared has increased from 5,72 percent to 6,49 percent. The results of this regression analysis are comparable to the results of the previous regression analysis. First it shows that there is a significant negative relationship between the reporting location of other comprehensive income and stock return: hypothesis 2b can be accepted again. Similar to regression 3 (shown in table 8), there is a significant negative relationship between the volatility of the income shown in the performance statement and stock returns. Also, similar to regression 5 (shown in table 11) there are significant positive relationships between ‘MKTSEC’ and ‘STOCK RETURN’ and ‘LEVERAGE’ and ‘STOCK RETURN’.
Table12. Regression results for regression 6 with robust standard errors (Dependent Variable = STOCK RETURN)
Variable Coef. Std. Err. t P>t
Vol.PS -1.466.474 .8285407 -1.77 0.077** LOCATION -.1324563 .0453672 -2.92 0.004*** OCI .2007019 2.01675 0.10 0.921 NI .2140462 .2560728 0.84 0.404 Vol.OCI -.00032 .0012959 -0.25 0.805 MKTSEC .0939956 .0429277 2.19 0.029** PENSION -.0320815 .0444191 -0.72 0.470 FORCUR .0452098 .0371677 1.22 0.224 LEVERAGE .454052 .1817895 2.50 0.013** LSIZE -.0255821 .0216744 -1.18 0.238 Cons. .2082799 .1356199 1.54 0.125 N 573 R-squared 0,0649
6. Results for testing H3a and H3b
6.1 Preliminary tests
Similar to the model which is used to test for hypothesis 2a and 2b, tests for heteroskedasticity, linearity and multicollinearity are performed for this model. The p-value of the White-test is equal to 0,0231 which indicates that there is a situation of heteroskedasticity. Similar to the previous regression analyses, regression analyses with robust standard errors are performed. These regressions adjust for the effect of heteroskedasticity. Second the Ramsey RESET test regarding linearity is performed. The p-value of this test is equal to 36.28 percent, which means that there is no situation of misspecification. Finally the test regarding multicollinearity is performed. The vif-value is equal to 1.12 which means that there is no situation of multicollinearity.
6.2 Descriptive statistics
Table 13 shows the descriptive statistics for the regression analyses used to test for the third pair of hypotheses (3a and 3b). The sample consists of the sample selection of both 2000 and 2012 which includes 573 observations. The dependent variables for this regression are the volatility of the income shown in the performance statement (hypothesis 3a) and stock returns (hypothesis 3b). Among other variables the most important independent variable is ‘YEAR’; this variable shows if there are significant differences between the dependent variables between 2000 and 2012. This dependent variable shows if the change in regulation in 2011 had any influence on stock returns or the volatility of the income shown in the performance statement. Out of the 573 observations, 311 observations (54,28%) are from 2012 where the other 262 observations (45,72%) are from 2000. The average volatility of the income shown in the performance statement is equal to 0,20926, while the average stock return was equal to 0,1245662. This indicates an average stock return of 12,46 percent. The mean of other comprehensive income scaled by total assets is equal to -0,0063472 while the mean of net income scaled by total assets is equal to 0,0617568. The mean of the volatility of other comprehensive income of this sample is equal to 1,868317 which means that the volatility of comprehensive income is on average nearly 1,9 times as high as the volatility of net income.
Table 13. Descriptive Statistics of the sample used to test for H3a and H3b
Panel A: Sample composition
Method Number of observations Percentage
2000 - OCI reported on balance sheet 255 44,50% 2000 - OCI reported in profit and loss 7 1,22%
2012 311 54,28%
Panel B: Summary statistics
Variable 1st Quartile Median 3rd Quartile Mean Standard Deviation Vol.PS 0,0054215 0,0129231 0,0259193 0,020926 0,0260201 STOCK RETURN -0,0869565 0,118021 0,2721335 0,1245662 0,4243378 YEAR 0 0 1 .4686411 .5233393 OCI -0,0068012 -0,0008098 0,0018145 -0,0063472 0,0635298 NI 0,0199043 0,0535827 0,0935323 .0617568 0,0646124 Vol.OCI 0,9510624 1,056528 1,461831 1,868317 4,443391 MKTSEC 0 0 1 0,2560976 0,436857 PENSION 0 0 1 0,3222997 0,4677647 FORCUR 0 0,5 1 0,5 0,5004361 LEVERAGE 0,0902196 0,1876894 0,2986433 0,2048804 0,1461806 LSIZE 5,003344 5,431693 5,832892 5,261902 0,9648712
In table 14 the results of a Spearman correlation test are shown. Similar to the previous tests, the volatility of the income presented in the performance statement is added to the correlation test. Compared to the Spearman correlation test performed about the sample of 2000 there is no correlation anymore between the variables ‘Vol.PS’, ‘Vol.OCI’ and ‘NI’. Furthermore, similar to the previous correlation test, there are relationships between size of the firms, leverage and the different components of other comprehensive income (FORCUR, PENSION and MKTSEC). This could be the case because bigger firms should have more resources which could be invested in foreign currency transactions, pensions and investments in available-for-sale securities. Furthermore there are several relationship between ‘YEAR’ and another variable. This implies that when firms report other comprehensive income in a statement of profit and loss report less amounts of other comprehensive income, bigger amounts of net income, spent less on pension expenditures, have lower amounts of debt and are smaller.
Table 14. Correlation of variables used for tests of H3a and H3b
Vol.PS YEAR OCI NI Vol.OCI MKTSEC PENSION FORCUR LEVERAGE LSIZE Vol.PS 1 P-value 0 YEAR 0,1079* 1 P-value 0,0097 0 OCI -0.0682 -0,816* 1 P-value 0.1028 0,0507 0 NI 0.0444 0,0944** 0.0430 1 P-value 0.2879 0,0237 0.3038 0 Vol.OCI -0.0381 -0,0167 -0.2228*** 0,0083 1 P-value 0.3627 0,6896 0.0000 0,8436 0 MKTSEC -0.0577 0,039 0.0803* -0,0545 0,0671 1 P-value 0.1674 0,3509 0.0544 0,1923 0,1081 0 PENSION -0.0437 -0,4042*** 0.0710* -0,1080*** 0,0041 0,0053 1 P-value 0.2961 0.000 0.0891 0,0096 0,9212 0,8990 0 FORCUR -0.0505 -0,01 0.0877** 0,0265 0,0465 0,0918** -0,1305*** 1 P-value 0.2272 0,8111 0.0357 0,5258 0,2665 0,0279 0,0017 0 LEVERAGE -0.0420 -0,0943** 0.0112 -0,1121*** -0,0928** -0,1967*** 0,0706* -0,0795* 1 P-value 0.3156 0,0238 0.7887 0,0072 0,0262 0,000 0,0912 0,0570 0 LSIZE -0.1705* -0,319*** 0.0240 -0,1362*** 0,0013 0,0974** 0,1874*** -0,0668 0,0090 1 P-value 0.0000 0.000 0.5658 0,0011 0,9751 0,0197 0,000 0,1104 0,8290 0
Notes: *** significant at 1 percent level; ** significant at 5 percent level; * significant at 10 percent level
6.3 Regression results
The results of the seventh regression analysis are shown in table 15. Again regression analyses with robust standard errors are performed; the significant results did not change compared to the regular regression analyses. In this regression analysis the possible effect of the obligated reporting location of other comprehensive income on the volatility of the income presented in a performance statement is tested (Hypothesis 3a) by measuring the effect of the reporting year. Equation 3 is used for this model. At first, there is no difference between the volatility of the income shown in the performance statement between 2000 and 2012 since the variable ‘YEAR’ is not significant (p-value = 0,197). This means that the change in SFAS 130 did not influence the volatility of the income presented in the performance statement and that hypothesis 3a should be rejected. Regarding the other variables, there is only one variable that shows a significant (negative) relationship with Vol.PS. This variable is the size of the firm. This implies that when a firm is bigger the volatility of the income presented in the performance statement is lower.
