MSc Accountancy & Control, variant Accountancy Faculty of Economics and Business, University of Amsterdam
The Effects of Earnings Volatility on Lobbying
Activities towards IFRS 9
Final Version of Master Thesis Student Name: Hu Xinpei Student Number: 10603468 Date: 2014-07-21
First Supervisor: Dr. Sanjay Bissessur Second Supervisor: Dr.Ir.Sander van Triest
Content
1. Introduction ... 4
2. Literature review and Hypothesis ... 6
2.1 Earnings volatility ... 6
2.2 Standard-setting and constituents ... 9
2.3 IAS 39 vs. IFRS 9 ... 12 2.4 Hypothesis development ... 17 3. Research design... 19 3.1 Methodology ... 19 3.2 Sample ... 20 4. Results ... 26 4.1 Hypothesis test ... 26 4.2 Robustness test ... 31 5. Conclusion ... 35 References ... 37 2
Abstract
This study investigates whether earnings volatility acts as a motive for lobbying activities towards IFRS 9 Financial Instruments. Lobbying banks comment to Exposure draft 2012/11 in current stage of the IASB due process and non-lobbying banks are employed as research samples. Using a binary logit model on relative data from 1998 to 2013 and controlling several variables, the results shows significant negative relationship between earnings volatility and lobbying behaviour (e.g. writing a comment letter). Thus, banks with lower earnings volatility before IFRS 9 amendments are made tend to lobby more. This finding is different from previous conclusions regarding distinguishes of earnings volatility characteristics between lobbyists and non-lobbying firms. Meanwhile, banks with higher leverage ratio, larger size lobby more. This correlates with prior literature. The study contributes to literature on constituents’ characteristics especially preparers’ characteristics and literature on lobbying analysis under specific IFRS (IFRS 9 in this study).
Key words: IFRS 9; Standard-setting; Earnings volatility; Lobbying.
1. Introduction
The International Accounting Standards Board (IASB) is formally operated since 2001 as the successor to the International Accounting Standards Committee (IASC). It is an independent standard-setting body overseen by the IFRS Foundation. The International Financial Reporting Standards (IFRS) are set and issued as development of a set of understandable and globally accepted standards (ifrs.org). The EU approved adoption of IFRS for EU-listed companies in 2002 and mandated the adoption regarding consolidated financial statements in 2005. There are countries outside Europe adopting IFRS, for instance, Brazil, Canada and South Korea (Jorissen, 2013). Currently, more than 100 jurisdictions are using IFRS, without including those planning to adopt IFRS (ifrs.org).
As the importance and influence of the IASB increase worldwide, it demands stronger legitimacy from its constituents. Legitimacy is a positive judgement by audience of an organisation and enhances its credibility (Schuman, 1995). Participation of constituents is seen as a key factor of legitimacy (Richardson and Eberlein, 2011). Therefore the due process that ensuring public participation of stakeholders becomes indispensable procedure to guarantee the success of the IASB. The due process is similar to which is established by the US Financial Accounting Standards Board (FASB). Interested parties such as preparers, accounting profession, regulators, and users from all geographical regions are welcomed to respond (Larson, 2007). Ways for constituents to involve standard-setting include submitting comment letters, participating in relevant meetings, and raising questions in public hearings (Georgiou, 2010; IFRSF, 2011). All comment letters are available at the IASB website.
This study focuses specifically at IFRS 9 Financial Instruments and its constituents’ participation. Additionally, I focus only on responses from preparers especially banks, as preparers operate in various industries while IFRS 9 is directly related to banking industry with its financial nature.
The IFRS 9 project (replacement of IAS 39) is divided into three phases by the IASB: I. Classification and measurement; II. Impairment methodology; III. Hedge accounting. Three discussion papers and five exposure drafts have been issued since the start of the IFRS 9 project. Looking at phase I classification and measurement, the related IFRS 9 is first published in 2009. I look into the latest exposure draft - Classification and Measurement: Limited Amendments to IFRS 9 (Proposed amendments to IFRS 9 (2010)) - under the first
phase issued in 2012 and select banks writing comment letters to this exposure draft as samples of research.
Previous researches on lobbying behaviour towards international standard-setting are rich within U.S. literature (e.g. Francis, 1987; Johnston, 2006; Gipper et al., 2013). However, studies investigating lobbying behaviours under the IASB’s standard-setting process are limited. Jorissen (2012) compares participation of preparers and users regarding 5 discussion papers and 28 exposure drafts issued by the IASB. Larson (2013) focuses on geographic diversity of constituents from 2001 to 2008. Specific to each standard of IFRS, Giner (2012) examine characteristics of constituents towards IFRS 2 Share-based Payments and Huian (2013) towards first phase of IFRS 9 Financial Instrument. Both of the two assess the IASB’s reaction to comments received in the due process. Kosi (2014) looks into determinants of lobbying behaviour towards IFRS 4 Insurance Contracts.
This study investigates differences of earnings volatility between lobbyists and non-lobbying banks. Earnings volatility acts as an important factor considered by both management and investors (Trueman, 1988; Gaver et al., 1995; Barnea, 1976). Higher earnings volatility threats quality of earnings (Xu, 2012). Fiechter (2011a) examines how the fair value option proposed in IAS 39 affect banks’ earnings volatility. As a reference, changes in accounting standards would affect a firm’s earnings volatility. Furthermore, if a firm perceive that the amendments of standards would increase volatility in its earnings figures, it will conduct lobbying activities to avoid or mitigate the foreseeable loss caused by those amendments. Therefore, I consider change in earnings volatility as a motive for lobbying and predict that firms of lower earnings volatility prior to the amendments of standards (IFRS 9 in this study) would lobby more.
I employ a logit model to include both lobbying and non-lobbying samples in time period from 1998 to 2013. Control variables include SIZE, ROA, YR_Dum, etc. The achieved results support the prediction, showing that earnings volatility is lower for lobbying banks than non-lobbying banks. And the leverage ratio is higher among lobbying banks compared to non-lobbying ones. Research results contribute to literature investigating how the earnings volatility distinguishes between lobbying firms and non-lobbying firms. Previous empirical results reach varied conclusions (Ang, 2000; Kosi, 2014) and this study presents a negative relationship between earnings volatility and lobbying behaviours. Moreover, this is the first research analysing constituents’ responses under current stage of
IFRS 9 project. The research content is quite advanced as the final IFRS 9 about classification and measurement has not been settled yet. Currently, researchers could not examine how the IASB reacts to these received comments but can only look into characteristics of constituents.
The study is organised as follows: Section 2 recaps important literatures on earnings volatility, theories of standard-setting and constituents’ involvement and elaborates on the replacement to IFRS 9, as well as newly raised amendments. Hypothesises are raised at the end of this section. Section 3 explains the model and sample used for the research. In section 4 the empirical results are presented and analysed in detail. Section 5 gives a short summary of the study and puts forward several future research points.
2. Literature review and Hypothesis
This section will look further into previous studies and is divided into four parts. The first part reviews why earnings volatility is important, its determinants, and some effects of standard setting on earnings volatility. The second part clarifies the role and process of standard-setting as well as participation of constituents within the process. The third part identifies the alteration of accounting standards from IAS 39 to present IFRS 9, in which more attention is paid on classification and measurement. Following the elaboration of prior literature in the parts mentioned above, hypothesis of this study are raised at the end of the section.
2.1 Earnings volatility
Earnings are the summary measure of firm performance achieved using accrual basis of accounting and are used by a wide range of users (Dechow, 1994). Earnings volatility indicates the extent of earnings of a corporation is stable or unstable. And analysts usually focus on figures of annual or quarterly earnings.
From managers’ perspective of views, unstable earnings decrease the quality of earnings prediction. For instance, when managers carry out financing activities for long-term investment, the predicted cash flows may not reliable. On one hand, the borrowing cost increases. On the other hand, the potential advocacy may lead to seizure of assets offered by lenders and even bankruptcy (Trueman, 1988). Management fights to create a stable earnings stream through income smoothing to sustain their compensation tied to accounting outcomes
(Lambert, 1984; Gaver et al., 1995) as well as to provide more ‘beautiful’ figures for estimation and evaluations of the firm’s earnings variability.
From investors’ perspective of views, a firm with earnings fluctuating heavily equals a more risky investment, which is also unfavourable to achieve desirable predictions of a firm’s future cash flows (Barnea, 1976). Gordon (1964) mentions that management will smooth reported income within the permission of accounting rules because investors relate stability to a higher dividend rate and a higher stock price. Moreover, earnings volatility is a vital element in relation to credit ratings (Ashbaugh-Skaife et al., 2006; Hann et al., 2007). Rating agencies regard financial statements as crucial source to examine the volatility of a company’s operation.
Earnings volatility is considered as one component of earnings quality (Xu, 2012). As the timing and matching problems induced by cash flows recognition, accruals are introduced to mitigate such issues, making earnings figures better represent firm performance (Dechow, 1994). The accrual quality measurement in Dechow (2002) is highly correlated with earnings volatility, while Wysocki (2005) note that most vital driver of this accrual quality measurement is the contemporaneous correlation between cash flows and accruals. To look further into how the fluctuations of earnings is caused, an earnings model is established, where earnings equal the total of cash flows and accruals (Dechow, 2002; Chen, 2008):
E = CF + ACC.1
Variance is often used to evaluate volatility. It calculates to what extent an individual variable deviate from the average amount of the variable. The volatility could then be quantified. After taking variance of the equation above, a new one can be achieved:
Var(E) = Var(CF) + Var (ACC) + 2
ρ
CF,ACC�𝑉𝑉𝑉𝑉𝑉𝑉(𝐶𝐶𝐶𝐶)�𝑉𝑉𝑉𝑉𝑉𝑉(𝐴𝐴𝐶𝐶𝐶𝐶).The Var(E) represents variance of earnings, Var(CF) represents variance of cash flows, Var(ACC) represents variance of accruals, and ρCF,ACC is the correlation coefficient of cash flows and accruals.
Earnings can be broken into pre-managed economic earnings and non-economic earnings according to literature on income smoothing mechanisms (Subramanyam, 1996; Graham et al. 2005). For the component of pre-managed economic earnings, the operating
1
E represents earnings, CF represents cash flows, and ACC refers to accruals.
7
cash flows can serve as a proxy as they are less limited to manipulation (Francis, 2004; Irvine and Pontiff, 2005). In Chen 2008, they denote �𝑉𝑉𝑉𝑉𝑉𝑉(𝐶𝐶𝐶𝐶) as operating volatility or economic volatility which reflects operating activities. The �𝑉𝑉𝑉𝑉𝑉𝑉(𝐴𝐴𝐶𝐶𝐶𝐶) proxies financial reporting volatility. The more volatile the accruals are, there would be greater frequency of transfers between earnings and cash flows. Opinions are unanimous that
ρ
CF,ACC is a measure forearnings measurement and captures the degree of earnings management. In Dechow (1994) the value of
ρ
CF,ACC is negative. Firms that manage earnings tend to have accruals negativelycorrelated with cash flow. However, such indication for earnings quality is controversial (e.g. Dechow, 1994; Leuz, 2003). In summary, earnings volatility is identified as the sum of operating volatility, financial reporting volatility and degree of earnings management.
For exploration of the determinants of earnings volatility, results show that all these three affect significantly (Chen, 2008). Further ranking indicates that operating volatility acts as the most vital contributor, followed by earnings management and financial reporting volatility.
There are also studies focusing specifically on the value of smooth cash flows, investigating the effects of operating volatility in firm value (e.g. Rountree 2008). Rountree (2008) find that the magnitude of effect is always large: when cash flows volatility of a firm increases by 1%, an approximate 0.15% reduction in firm value appears. In addition, they explore whether market distinguishes the streams of cash flows volatility and financial reporting volatility identified in Chen (2008)’s model. Their results show that investors tend to only care about cash-flow volatility regardless of smoothing activities under the existing of accrual flexibility. One reason is that smoothing through accruals indicates manager’s estimation of future cash flows, existence of measurement error and potential manipulation. Thus in what way a firm mitigates earnings volatility and achieves smoothing matters a great deal.
Considering the effects of externals, changing of accounting standards could have certain impact on a company’s earnings stream. For instance, many studies examine the employment of fair value accounting and its effects on earnings volatility (Barth, 1995; Hodder 2006; Duh et al., 2012). Fair value accounting is addressed to better reflect the underlying economic values than historical accounting, under which changes in values are recognised in net income although not realized, leading to more volatile earnings (Barth,
1995). Duh et al. (2012) examine the adoption of IAS 39 on earnings volatility as IAS 39 requests the recognition level of unrealized gain and loss under fair value accounting within a larger scope of financial and derivative-financial instruments. In addition, they note that the employment of either hedge accounting or the fair value option could reduce mixed-measurement volatility under IAS 39 which arises because different valuation methods are used on financial assets and liabilities, thus making the volatile earnings less ‘artificial’. As verification, Fiechter (2011a) achieves that banks adopting fair value option in order to reduce accounting mismatches show lower earnings volatility and the FVO tends to be more efficient in decreasing mixed-measurement volatility than hedge accounting. The results support IASB’s introducing of such optional fair value accounting method.
2.2 Standard-setting and constituents
The political character of the standard-setting process has been highlighted in plenty of previous researches (e.g. Fogarty, 1994; Stenka and Taylor, 2010). Setting accounting standard is identified as a political action (Solomons, 1978). Gipper et al. (2013) define the political influence as deliberate intervention in the standard-setting process conducted by an economic entity in order to increase its economic value, wealth, or to realise other beneficial goals that are not in line with standard-setter’s mission. Thus, conflicts are foreseeable and unavoidable when setting the accounting standards as public resources are rather limited.
Elias (2011) argues that politics and power are two essential drivers of accounting standard-setting process and outcomes. Power, as defined by Parsons (1963), is “the generalized capacity to secure the performance of binding obligations”. The capacity is attributed to actors and external actors contain all the interested parties (Kwok and Sharp, 2005). Exercising power to influence accounting standards relates to allocation of resources. Although ways and results of exercising power are different, chasing for benefits in the trade-offs during the process it always an incentive (Parsons, 1963) and is regarded as winning and losing matters in Amershi et al. (1982). The role of economic consequences is also highlighted (Fogarty, 1994). Economic consequences would happen if changes of accounting standard lead to changes in a firm value or wealth of whom using the figures to make decision or affected by such decisions (Holthausen, 1983; Brüggemann, 2013). With the acknowledgement that economic consequences could be a potential driving factor for the standard-setting, distribution of such consequences as part of trade-offs during the process would be meaningful.
Institutional theory serves as an explanation that organizations respond to social expectations (Fogarty, 1992; Oliver, 1991). Organizations operate in a social environment of cultural-cognitive, norms and regulates with predictable sequences of action and reaction (Oliver, 1991). Within such system the organisations behave to meet social expectations beyond basic requirements. Regulators such as the IASB consider the position of relative parties when setting accounting standards and thus can sustain themselves.
Legitimacy theory is more frequently used when analysing the standard-setting process (Durocher et al., 2007; Richardson and Eberlein, 2011; Jorissen 2013; Larson 2013). Legitimacy refers to the approved result by an audience regarding the appropriateness of an organization and its behaviours (Richardson and Eberlein, 2011). Constituents are key audience that determine an organization’s survival (Larson, 2013). The IASB as an international private sector standard setter, receiving response from all stakeholders is essential to know about their views (Georgiou, 2002; Jorissen, 2013), mitigate conflicts (Kwok and Sharp, 2005) and lead to a better compliance with the standards (Larson, 2013). Thus, constituents’ involvement is seen as fundamental to the IASB’s legitimacy. Since the first adoption of IFRS in Europe in 2002 and the mandatory adoption of IFRS in EU in 2005, there have been 130 countries adopting IFRS. Such global approval relies on the standard-setting legitimacy of the IASB (Jorissen, 2013).
To ensure the involvement and lobbying activities of constituents, a due process is employed by the IASB, enabling interest parties lobby before the adoption of the IFRS. Six stages constitute the IASB’s due process (IFRSF, 2011):
Stage 1: Setting the agenda. Issues are raised or reviewed when the users’ needs are demanded or the IASB’s Conceptual Framework for Financial Reporting is revised. A majority vote at the IASB meeting for certain agenda items would approve the adding. Stage 2: Project planning. In this stage, the IASB decides whether to carry out the project jointly with other standard-setter (e.g. the FASB). A project team is established and the plan for the project is drawn up. Stage 3: Development and publication of a discussion paper. A discussion paper (DP) is not mandatory while the IASB normally publishes it to clarify the agenda issue and call for comments from interest parties. The comment period is 120 days or longer and the comment letters are posted on website. Further comments may be needed in form of public hearings, round-table meetings, etc. Stage 4: Development and publication of an exposure draft. An exposure draft (ED) is mandatory and is in the form of a proposed
IFRS or amendment to an IFRS. It invites comments on draft IFRS or draft amendments and is a main source for consulting public. The following procedures are nearly the same as Stage 3 except an additional requirement that the IASB must consult the IFRS Advisory Council and keep the contact with constituents. Stage 5: Development and publication of an IFRS. If one ED is considered not sufficient to solve the issue, a second ED will be published to further the revision following the same process in stage 4. As the completion of project can be seen, the IASB prepares feedback statement and examines effects of the upcoming IFRS. After the new draft or revision is satisfied, balloting and final publication of IFRS is carried out. Stage 6: Procedures after an IFRS is issued. Staff of the IASB meet interested parties at a regular basis to follow the application of new IFRS or the amendments. Educational events are held by the IFRS Foundation to supplement the implementation. Review activities are carried out by the IASB normally after two years from the revised standards being adopted considering unexpected costs or other problems occurred.
According to the interest group theory, an industry operates in relation to various interest groups (constituents) and these interest groups will lobby towards various amounts and types of regulation (Scott, 2009). Lobbying behaviour in the standard-setting process can be defined as individual or organisation efforts to affect or obstruct the proposed standards (Watts and Zimmerman, 1978; Weetman et al., 1996; Georgiou, 2004; Durocher et al., 2007). Interested parties will spare efforts to involve and affect the final standard on basis of their potential economic benefits (Watts and Zimmerman, 1978). They must decide whether to lobby, when to take action, and what exactly to argue (Giner, 2012). Tokar (2005) notes that international accounting standard-setters have more diverse constituents than national standard-setters including investors, international accountancy firms, preparers, the international community (e.g. the World Bank), trade unions, creditors and suppliers, etc., and the potential economic consequences of proposed standards would be worldwide.
There are plenty of researches in the lobbying literature. However, the amount of researches regarding the participation of constituents in the IASB’s due process described above is rather limited. While we can still identify two main streams among these studies (Kenny and Larson, 1993). One stream focuses on either or both of the characteristics of lobbyists and determinants of lobbying behavior (Georgiou, 2010; Jorissen, 2013; Larson and Herz, 2013; Kosi, 2014), the other stream examines how the IASB reacts to voices from constituents (Kwok and Sharp, 2005; Giner, 2012; Huian, 2013).
Formal lobbying methods include submitting comment letters in response to the IASB’s consulting invitation, participating as a consultant in the project group, participating in the IASB’s round-table meetings, speaking at public hearings, etc. (Georgiou, 2010; Orens et al., 2011) There are also indirect lobbying methods (Sutton, 1984), for instance, submitting comments to Accounting Standard Council (ASC) and the European Financial Reporting Advisory Group (EFRAG). Among the participation methods, comment letters are regarded as a good proxy for analysing lobbying activities (Georgiou, 2004) and are frequently used because they are public available, rich in contents (Stenka and Taylor, 2010), and have fatal impact on final standard (Hansen, 2011).
For studies examining constituents’ characters, Georgiou (2010) indicates that users of financial statements are more likely not to lobby in the due process compared to preparers. They gain less benefits and bear high costs in lobbying, and the possibility of success is low (Sutton, 1984; Weetman et al., 1996). Larson and Herz (2013) conduct their research on geographic level. They find countries with EU membership, donations to the IASB, and developed equity market lobby more. For the determinants of lobbying behaviour, Kosi (2014)’s research regarding the replacement process of IFRS 4 attributes income volatility as a driver to lobby because firms will be against the increased uncertainty if the proposed standards would cause such economic consequence.
For the second research stream investigating the IASB’s reaction to lobbying, Giner (2012) focusing on IFRS 2 shows that preparers response more and earlier to the IASB than other interest groups, while none of the interested parties had a dominant effect. Huian (2013) analyses the involvement of major interested groups in early stage of IFRS 9. Findings present the strongly objected issues were revised or significantly amended, confirming that comment letters are influential method of lobbying.
2.3 IAS 39 vs. IFRS 9 Replacing the IAS 39
The financial crisis occurred in the second half of 2008 led to a worldwide rethinking on the present standards regarding recognition and measurement of financial instruments (Paananen et al., 2012). The IAS 39 - standard on the recognition and measurement of financial instruments – has caused much confusion and heated debates (Fiechter, 2011b). The current
IAS 39 categorises financial assets into four streams (Figure 1), with two sub-streams under fair value through profit or loss (FVTPL).
Figure 1: IAS 39 classification of financial assets
Loans and receivables are amounts that the creditor expects debtor to repay at a specific date. A held-to-maturity investment is a non-derivative financial asset that has fixed or determinable payments and a maturity period, and an entity has both ability and intention to hold it to maturity. And available-for-sale distinguishes those that are not expected to be held for a lengthy period of time or until maturity. Fair value measure requires a financial asset recognised at fair value on initial recognition and the subsequent changes in fair value attribute to profit or loss. This category (FVTPL) draws the most attention especially after the occurrence of 2008 financial crisis. Manager of financial institutions began to resist fair value accounting (Guerrera and Hughes, 2008). Politicians also involved as the emerging demand from financial institutions (André, 2009; Elias, 2011). The pressure leads the IASB amend IAS 39 in October 2008 in order to achieve fair comparison with the U.S. GAAP (Generally Accepted Accounting Principles). The revisions allow an entity to reclassify non-derivative financial assets out of the FVTPL and AFS under certain conditions (IASB, 2008), thus mitigating opposes to fair value accounting.
In order to further reduce complexity and increase relevance of the standards, the IASB has accelerated its project to replace IAS 39. The IFRS 9 Financial Instruments is supposed to eventually replace IAS 39 Financial Instruments: Recognition and Measurement. The project is jointly conducted by the IASB and FASB and is never an easy task. After the publication of discussion paper Reducing Complexity in Reporting Financial Instruments in 2008, the IASB split the replacement into three phases: I. Classification and measurement; II. Impairment methodology; III. Hedge accounting.
Currently, the Hedge accounting phase has been finalized. The IASB is continuing with some more limited amendments towards the classification and measurement requirements that are already included in IFRS 9 and is also discussing the expected credit loss impairment model to be included in IFRS 9. Once these are completed, a final version of IFRS 9 will be published including all the phases and a new mandatory effective date (IASB, 2013).
The IFRS 9 applies a new approach for all types of financial assets, including those that contain embedded derivative features instead of more complex requirements in IAS 39. The classification is based on an entity’s business model for managing its financial assets and the contractual cash flow characteristics of the financial asset. In addition, besides amortised cost and FVTPL, a new category named fair value through other comprehensive income (FVOCI) is employed. Fair value option is also applicable to this category (IASB 2009, 2012). Embedded derivatives are only separated from the host contract when the contract is not an asset within the scope of IFRS 9. Otherwise the entire hybrid contract is accounted for as one instrument. Rules about reclassification of financial assets are enriched in IFRS 9.
As to standards of financial liabilities, two measure categories of FVTPL and amortized cost remain. IFRS 9 (2010) requires that when a financial liability is designated as at fair value through profit or loss under the fair value option, the change in the fair value of the liability attributable to changes in the issuer’s own credit risk should be recognised in OCI. The amount presented in OCI due to changes of credit risk could not be recycled to profit or loss in subsequent period. Reclassification of financial liabilities is not allowed under IFRS 9. In 2012 it is proposed that an entity could early apply the own credit risk related paragraphs without early applying other requirements of IFRS 9.
Standards regarding hedge accounting are published in November 2013, proposing a new model for hedge accounting. The model aligns hedge accounting with risk management activities and improves the quality of disclosure. The disclosures will clarify the effect that hedge accounting has had on the financial statements, the risk management strategy, as well as derivatives that have been entered into and their effect on the entity’s future cash flows. More entities, especially non-financial institutions, could employ hedge accounting to reflect actual risk management activities (IASB, 2013). Meanwhile, users can better understand these activities and their effects.
Major changes
Key changes from the IAS 39 to the IFRS 9 regarding financial instrument are shown in the following table (Table 1):
Table 1
IAS 39 IFRS 9
Debt Instrument Classification
Fair value through profit and loss (FVTPL);
Loans and receivables (LAR); Held-to-maturity (HTM); Available-for-sale (AFS);
Amortised cost;
Fair value through profit and loss (FVTPL); Fair value through other comprehensive income (FVOCI); Classification
Basis
The entity’s intention (to hold till maturity, trading for short term profits, derivative, loan or receivable, or intentional designation);
Business model and the contractual cash flow characteristics;
Debt
Instruments Measurement
Measured at amortised cost if under HTM or as LAR; others are
measured at fair value.
See Figure 2.
Embedded derivatives
Embedded derivatives are
separated from the hybrid contract and are measured at FVTPL.
No bifurcation of asset. The financial asset is examined entirely based on the contractual cash flows; if any of the cash flows do not represent either payments of principal or interest, the entire asset is measured at FVTPL.
Fair value option
An entity could designate a financial asset to be measured at fair value at initial recognition without additional requirements.
A financial asset can be designated as FVTPL on initial recognition only if it eliminates or significantly reduces an accounting mismatch under amortised cost.
Classification and measurement under IFRS 9
There are several stages in this Phase. The first Exposure Draft Financial Instruments:
Classification and Measurement is issued in July 2009. The ED proposed that if an asset
earned predictable cash flows, then it could be measured at amortised cost. If returns of the asset were unpredictable, for instance a share portfolio or derivative, then it should be recognized under fair value measure. The European Commission (EC) was unpleasant with such revision and called for more changes (e.g. more flexible valuation standards) during the IASB’s consultation (Sanderson and Tait, 2009). The IASB published the first part of IFRS 9 in November 2009. Accordingly, it included many modifications raised by the EC. And the amendment of financial liabilities is put off until 2010.
However, the EC advisory body, European Financial Reporting Advisory Group (EFRAG), decided to delay the endorsement of the new standard. Reasons of the delay are considered to be concerns about some French, German and Italian banks with large investment banking activities (Elias, 2011). The amendments would force these banks to book substantial losses on their derivative portfolios. Subsequently, the IASB considered making modifications to IFRS 9 (2010) and issued one more Exposure Draft Classification
and Measurement: Limited Amendments to IFRS 9 (Proposed amendments to IFRS 9 (2010))
in November 2012. As this study focus on lobbying activities towards this ED (ED/2012), a further looking into these limited amendments is followed.
Consistent with FASB’s existing proposals, the IASB introduce a fair value through other comprehensive income (FVOCI) measurement category in 2012 ED. A financial asset shall be measured at FVOCI if it is with contractual cash flows that are solely principal and interest on the principal amount outstanding and is held in a business model in which assets are managed both in order to collect contractual cash flows and for sale (ED/2012/4.1.2A). The assessment of the business model is performed at an aggregated level (rather than the instrument level) (Ernst & Young, 2012). After satisfying that the solely payments of principal and interest test (SPPI test), financial assets with business model to collect contractual cash flows would be measured at amortised cost. Financial assets fail either the business model criteria for FVOCI or for amortised cost (e.g. portfolios held for trading or managed to realise cash flows through active buying and selling) would be measured at FVTPL. If an asset fails the SPPI test, it goes directly to the category of FVTPL (ED/2012/B4.1.8A). The procedure of deciding classification and measurement described in this paragraph is shown in Figure 2.
Figure 2
Note: The procedure of classification and measurement of a financial instrument proposed in the 2012 ED. (IASB, 2012. Snapshot: Financial Instruments: Classification and Measurement (Limited Amendments to
IFRS 9))
The FVOCI category subjects to the same impairment and income recognition models as financial assets measured at amortised cost (ED/2012/5.2.2). Cumulative fair value gains and losses recognised in OCI would be recycled to profit or loss upon derecognition (ED/2012/5.7.1A). If designating a financial asset to be measured at FVTPL eliminates or significantly reduces accounting mismatch, an entity could use this fair value option at initial recognition.2 This option extends to financial assets qualifies FVOCI (ED/2012/BC74). The new category means that some financial assets recognised under FVTPL would now be measured at FVOCI. Thus less volatility in profit or loss is expected.
2.4 Hypothesis development
Previous studies find preparers lobby more than other interested groups in the due process and are more incentive to involve (Georgiou, 2004; Giner 2012). This study focuses on lobbying behaviour of prepares.
The perceived negative impact on a company’s operation and accounting figures drives preparers’ participation (Francis, 1987; Ang, 2000; Jorissen, 2012). In addition, Elbannan and McKinley (2006) believe that an incremental uncertainty of accounting
2 Paragraph 4.5 of IFRS 9 (2009).
17
information drives a firm’s involvement. Based on the significance of smoothed earnings highlighted in the first part of this section and the prediction of Elbannan and Mckinley (2006), earnings volatility would serve as a driver for lobbying if the proposed standards increase the uncertainty of a firm’s earnings stream. Relevant studies can be found regarding the role of earnings volatility in firms’ lobbying (Ang, 2000; Jorissen, 2012; Kosi, 2014).
One frequently mentioned issue in comment letters regarding the 2011 ED is the solely payment of principal and interest test (SPPI test) is beyond identification of the business model. Before looking at an entity’s business model, a financial asset should first satisfy that its contractual cash flows are solely payment of principal and interest on the principal amount outstanding.3 Otherwise the instrument will be measured at FVTPL. It is argued that this requirement this will lead to some financial instruments disqualify from being measured at amortised cost and will be recognised under fair value measure (FVTPL). There seems to be no improves in relevance of such change. However, the volatility of earnings would increase because of the increased use of fair value accounting. According to the prediction attained in last paragraph, this major amendment in the 2011 ED affecting earnings volatility would drive preparers especially those have greater financial instruments diversity to lobby to the IASB.
I perceive that firms with lower earnings volatility are more sensitive to changes in earnings stream because their earnings are more stable and easier to reflect the influence of revised standards. Therefore they have higher motivation to participate in the due process to secure the smooth of earnings and mitigate the fluctuation under the proposed standards. Given all analysis, the first hypothesis of this study is made as follows:
H1: Firms lobbying in the IASB’s due process have lower earnings volatility compared to non-lobbying firms.
On the other hand, earnings volatility indicates higher risks to investors, lobbying activities of preparers could be read as firms do not want to bear such additional risks. I perceive that the lobbying firms have higher potential operating risks. Leverage ratio serves as a proxy for operating risk of a firm, and the incremental earnings volatility would increase firm’s cost of debt. Thus:
3
Detailed explanations of this amendment can be found in part three of this section: Classification and measurement under IFRS 9.
18
H2: Firms that lobbying have higher leverage ratio than those non-lobbying firms.
3. Research design 3.1 Methodology
Different from using content analysis of comment letters as a major research method, I establish a logit model on basis of prior empirical researches (Ang, 2000; Jorissen, 2012; Kosi 2014):
L = β0 + β1EVOL + β2SIZE + β3LEV + β4ROA + β5CFVOL + β6P/B + ε (1) Where:
L = Lobbying activities by a firm. It equals 1 if a comment letter is submitted in response to the ED and 0 if not;
EVOL = Earnings volatility measured at the standard deviation of quarterly earnings before tax scaled by total assets;4
SIZE = Natural logarithm of market capitalisation at the fiscal year-end. LEV = Total liabilities/ total assets;
ROA = Return on assets as net income before extraordinary items divided by total assets at fiscal year-end;
CFVOL = Cash flows volatility measured at standard deviation of CF divided by total assets;
P/B = Price to book ratio is measured at the stock price divided by the total equity. Lobbying (L) is the dependent variable in the regression model. It equals 1 if a company participates in the due process and is 0 is the company does not participate. To be specific, it does not write a comment letter to the IASB. Earnings volatility (EVOL) is scaled by total assets to evade the disturbing of size variations of samples and is calculated as the standard deviation of quarterly earnings before tax (Waymire, 1985; Fiechter, 2011a). Leverage (LEV) is used for testing H2 and controls for a firm’s financing structure. Jorissen (2012) find more profitable firms lobby more. They use ROA as the proxy for a firm’s profitability.
Based on that operating volatility ranks higher in the contribution to earnings volatility (Chen, 2008; Rountree, 2008), I set an additional proxy of cash flows volatility to investigate its influence on lobbying activities. Moreover, firms with higher P/B ratio imply
4
EARN = EBIT/Total assets; EARN_mean is sorted by company_ID; EARN_diffsq = (EARN-EARN_mean)^2; EVOL= ((l2.EARN_diffsq + l.EARN_diffsq + EARN_diffsq)/3)^0.5.
19
more growth opportunities and tend to be more sensitive to changes in accounting standards (Chen, 2008).
Further, to control the effects of years within time period of 1998 – 2013, year dummy variables are introduced and added in the equation (1). Thus, an adjusted model is generated:
L = β0 + β1EVOL + β2SIZE + β3LEV + β4ROA + β5CFVOL + β6P/B + β7YR_Dum2 + β8YR_Dum3 + β9YR_Dum4 + β10YR_Dum5 + β11YR_Dum6 + β12YR_Dum7 + β13YR_Dum8 + β14YR_Dum9 + β15YR_Dum10 + β16YR_Dum11 + β17YR_Dum12 + β18YR_Dum13 + β19YR_Dum14 + β20YR_Dum15 + β21YR_Dum16 + ε (2)
Where:
L = Lobbying activities by a firm. It equals 1 if a comment letter is submitted in response to the ED and 0 if not;
EVOL = Earnings volatility measured at the standard deviation of quarterly earnings before tax scaled by total assets;5
SIZE = Natural logarithm of market capitalisation at the fiscal year-end. LEV = Total liabilities/ total assets;
ROA = Return on assets as net income before extraordinary items divided by total assets at fiscal year-end;
CFVOL = Cash flows volatility measured at standard deviation of CF divided by total assets;
P/B = Price to book ratio is measured at the stock price divided by the total equity. YR_Dum2 = year effects during time period of 1998-2013. It equals 1 if the data are in year 1999 and 0 otherwise;
YR_Dum3 = It equals 1 if the data are of year 2000 and 0 otherwise; …
YR_Dum16 = It equals 1 if the data are of year 2013 and 0 otherwise; 3.2 Sample
Samples selection
5
EARN = EBIT/Total assets; EARN_mean is sorted by company_ID; EARN_diffsq = (EARN-EARN_mean)^2; EVOL= ((l2.EARN_diffsq + l.EARN_diffsq + EARN_diffsq)/3)^0.5.
20
The study focus on comment letters responding to the Exposure Draft Classification and
Measurement: Limited Amendments to IFRS 9 (Proposed amendments to IFRS 9 (2010))
(2012). As samples of lobbyists, the comment letters are collected from the IFRS website (ifrs.org). 169 comment letters have been received till the commenting period closed on 28 March 2013. 52 CLs are written by preparers, of which 32 CLs are from banks (Table 2, Panel A). After excluding those of which relative data could not be achieved in Datastream, 21 banks are selected as samples for banks conducting lobbying behaviour (Table 2, Panel B). An overview of geographic region of these banks is shown in Table 2.
Table 2 Overview of CLs
Panel A: Overview of all banks' CLs Country Number of CLs Germany 2 UK 4 Spain 1 Austria 1 France 4 Switzerland 2 Australia 3 Geographic region CLs Percentage Singapore 1 European 17 53% China 11 Asia 13 41%
Japan 1 North America 2 6%
USA 2 Oceania 3 10%
Total 32 Total 32 100%
Panel B: Overview of bank letters with reachable data Country Number of CLs Germany 2 UK 3 Spain 1 Austria 1 France 3 Geographic region CLs Percentage Switzerland 2 European 12 57% Australia 2 Asia 6 28%
China 6 North America 1 5%
USA 1 Oceania 2 10%
Total 21 Total 21 100%
Notes: Panel A provides country and regional distribution of bank CLs from IFRS website. Panel B provides such distribution of banks with available data.
As for samples of non-lobbying banks, I use available data from banks worldwide rather than only four of the continents presented in Table 2. For instance, Africa, South America. The time period of data from both lobbyists and non-lobbying banks is from 1998 to 2013. Considering together, the final amount of adopted observations is 4,439 after deleting missing data.
Descriptive statistics
Distributional statistics of lobbying and non-lobbying banks together and correlation coefficients between each two variables are shown in Table 3. I first combine the lobbying and non-lobbying samples together to show a uniformed statistics (Panel A) and then achieve lobbying (Panel B) and non-lobbying (Panel C) distributional statistics as what Ang (2000) does. Distribution of frequencies on financial years and countries is reported in Table 4.
Table 3
Descriptive Statistics
Pannel A: Distributional Statistics
Variable N Mean Std. Dev. Q1 Median Q3
LETTER 7,554 0.069 0.253 0.000 0.000 0.000 EVOL 5,000 0.01 0.024 0.003 0.005 0.010 SIZE 7,239 16.61 2.616 14.761 16.634 18.330 LEV 7,534 0.191 0.162 0.063 0.153 0.279 ROA 7,536 0.009 0.033 0.003 0.008 0.014 CFVOL 5,627 0.026 0.191 0.006 0.012 0.024 PB 7,234 2.098 17.997 0.874 1.380 2.094
Pannel B: Lobbying observations
Variable N Mean Std. Dev. Q1 Median Q3
LETTER 520 1.000 0.000 1.000 1.000 1.000 EVOL 433 0.005 0.004 0.002 0.004 0.005 SIZE 495 17.592 1.183 16.961 17.496 18.182 LEV 520 0.271 0.126 0.181 0.258 0.356 ROA 520 0.005 0.004 0.003 0.005 0.008 CFVOL 387 0.012 0.011 0.005 0.010 0.016 PB 495 1.673 1.004 0.913 1.473 2.233 22
Panel C: Non-lobbying observations
Variable N Mean Std. Dev. Q1 Median Q3
LETTER 7,034 0.000 0.000 0.000 0.000 0.000 EVOL 4,567 0.011 0.025 0.003 0.005 0.010 SIZE 6,744 16.538 2.677 14.623 16.469 18.344 LEV 7,014 0.185 0.163 0.058 0.142 0.267 ROA 7,016 0.009 0.035 0.004 0.008 0.014 CFVOL 5,240 0.028 0.198 0.006 0.013 0.025 PB 6,739 2.13 18.644 0.873 1.374 2.078
Pannel E: Correlation coefficients
Var. LETTER EVOL SIZE LEV ROA CFVOL PB
1 LETTER 1 -0.1225 0.1001 0.2020 -0.1468 -0.0782 0.0105 2 EVOL -0.0679 1 -0.1637 0.1917 0.1643 0.3662 0.0801 3 SIZE 0.0792 -0.0306** 1 -0.2441 -0.0150 0.0285* 0.1194 4 LEV 0.1575 0.0618 -0.2223 1 0.0258* 0.1857 0.1192 5 ROA -0.0500 -0.2307 0.0455 -0.0313** 1 0.2399 0.5214 6 CFVOL -0.0771 0.4918 0.0346** 0.0623 0.0211 1 0.1449 7 PB 0.0030 0.0272* 0.0986 0.0423 0.1771 0.0765 1
Notes: Panel A, Panel B and Panel C of Table 3 reports distributional statistics of sample used in the study. N refers to the frequencies; Q1, Median, Q3 are three-number summary using quartile measure. Descriptions of definitions for each variable are in the methodology section. Panel D shows correlation coefficients between each two variables. Above the diagonal are Spearman correlation coefficients. Below the diagonal are Pearson’s correlations. The boldfaced values are at 1% significant level. **, * are values significant at 5% and 10% level respectively.
Panel B and Panel C of Table 3 report descriptive statistics of lobbying and non-lobbying banks respectively. While as the sample pool for non-non-lobbying banks is much larger than the amount of lobbying banks, figures in Panel C are not significantly distinguished from those in Panel A. Focusing on EVOL and LEV, the mean of EVOL regarding lobbying banks is 0.005 with standard deviation of 0.004. The mean value considering non-lobbying banks is 0.011 and the standard deviation is 0.025. The values show a relatively concentrated low level of earnings volatility within lobbying banks, serving as potential evidence that banks conduct lobbying are of lower earnings volatility. Meanwhile, mean value of LEV is 0.271 in Panel B with a standard deviation of 0.126, higher than the number of 0.185 in Panel C. Comparison between median amount of 0.258 and 0.142 also correlates with a positive relationship between LEV and Lobbying.
Looking at Spearman correlations above the diagonal in Panel D (Table 3), most of the values show significance. Only two of the numbers are insignificant, two are at 10% significant level, and the remaining values are all significant at 1% level. Values of -0.1225 regarding EVOL and 0.2020 regarding LEV in first line should be highlighted. The coefficients show a negative relationship between EVOL and LETTER and a positive relationship between LEV and LETTER, acting as implications for the correctness of predicted signs raised in the hypothesis.
Table 4 presents frequency statistics of selected sample on both financial years and countries. Data are collected from year 1998 to 2013 and from banks of 63 countries all over the world. The origin sources are 7,554 frequencies in total and are reflected in Table 4. The amount of available observations shows an increasing trend since year 1998, excepting a bit decrease in frequencies within year 2013. Japan and United States rank highest on percentage of all the frequencies with India, Switzerland and Italy followed. Having either more banks listed on market or a more regulated reporting atmosphere could contribute to more achievable financial data.
Table 4
Frequency Statistics
Panel A: Frequencies of financial years
Fyearend Freq. Percent Cum.
1998 330 4.37 4.37 1999 343 4.54 8.91 2000 366 4.85 13.75 2001 381 5.04 18.80 2002 403 5.33 24.13 2003 461 6.10 30.24 2004 480 6.53 36.59 2005 511 6.76 43.35 2006 528 6.99 50.34 2007 536 7.10 57.44 2008 543 7.19 64.63 2009 549 7.27 71.90 2010 552 7.31 79.20 2011 555 7.35 86.55 2012 554 7.33 93.88 2013 462 6.12 100.00 Total 7,554 100.00 24
Panel B: Frequencies of countries
Country Freq. Percent Cum.
ABU DHABI 134 1.77 1.77 ARGENTINA 98 1.30 3.07 AUSTRALIA 112 1.48 4.55 AUSTRIA 107 1.42 5.97 BAHRAIN 133 1.76 7.73 BELGIUM 43 0.57 8.30 BRAZIL 100 1.32 9.62 BULGARIA 32 0.42 10.05 CANADA 128 1.69 11.74 CHILE 90 1.19 12.93 CHINA 95 1.26 14.19 COLOMBIA 95 1.26 15.45 CROATIA 25 0.33 15.78 CYPRUS 16 0.21 15.99 CZECH REPUBLIC 16 0.21 16.20 DENMARK 80 1.06 17.26 DUBAI 49 0.65 17.91 EGYPT 74 0.98 18.89 FINLAND 30 0.40 19.29 FRANCE 177 2.34 21.63 GERMANY 139 1.84 23.47 GREECE 112 1.48 24.95 HONG KONG 77 1.02 25.97 HUNGARY 16 0.21 26.18 INDIA 317 4.20 30.38 INDONESIA 128 1.69 32.08 IRELAND 16 0.21 32.29 ISRAEL 93 1.23 33.52 ITALY 266 3.52 37.04 JAPAN 881 11.66 48.70 JORDAN 138 1.83 50.53 KUWAIT 99 1.31 51.84 LUXEMBOURG 15 0.20 52.04 MALAYSIA 160 2.12 54.16 MALTA 32 0.42 54.58 MEXICO 56 0.74 55.32 MOROCCO 82 1.09 56.41 NETHERLANDS 31 0.41 56.82 NIGERIA 121 1.60 58.42 NORWAY 48 0.64 59.05 OMAN 49 0.65 59.70 PAKISTAN 123 1.63 61.33 PERU 81 1.07 62.40 25
PHILIPPINES 139 1.84 64.24 POLAND 182 2.41 66.65 PORTUGAL 80 1.06 67.71 QATAR 78 1.03 68.75 ROMANIA 25 0.33 69.08 RUSSIAN FEDERATION 52 0.69 69.76 SINGAPORE 48 0.64 70.40 SLOVENIA 16 0.21 70.61 SOUTH AFRICA 92 1.22 71.83 SOUTH KOREA 67 0.89 72.72 SPAIN 129 1.71 74.42 SRI LANKA 101 1.34 75.76 SWEDEN 64 0.85 76.61 SWITZERLAND 306 4.05 80.66 TAIWAN 118 1.56 82.22 THAILAND 112 1.48 83.70 TURKEY 143 1.89 85.60 UNITED KINGDOM 192 2.54 88.14 UNITED STATES 765 10.13 98.27 VENEZUELA 131 1.73 100.00 Total 7,554 100.00
Notes: This table shows frequencies distributions regarding financial years and countries respectively. The percentage of frequencies in each year or country is shown in the third column. The cumulated percentage is presented in the fourth column.
4. Results
This section is divided into three parts. Firstly, the univariate test among variables employed is performed. Secondly, hypothesis raised in previous section are tested using the probit model. Thirdly, robustness of the research results is examined with setting additional constraints on sample selection.
4.1 Hypothesis test Univariate results
Based on the clarification of figures in Table 3 Descriptive statistics, I run an additional univariate test. The results using non-lobbying and lobbying samples are as follows in Table 5. Mean and median figures are shown in sections A and B. Section C presents t-statistic results for the H0: Mean (A) = Mean (B) and z-statistics from two-sample Wilcoxon
sum test. The two tests regarding EVOL and LEV both show great significance, confirming the raised hypothesises from the uniavriate perspective.
Table 5 Univariate results Section A Non-lobbying observations EVOL: n = 4567 LEV: n = 7014 Section B Lobbying observations EVOL: n = 433 LEV: n = 520 Section C Test of H0: A = B Mean Median Std.
Dev. Mean Median
Std. Dev. t-statistics (p-value) z-statistics (p-value) EVOL 0.011 0.005 0.025 0.005 0.004 0.004 5.123 9.283 0.000 0.000 LEV 0.185 0.142 0.163 0.271 0.258 0.126 -11.783 -15.198 0.000 0.000
Notes: T-statistics are from tests on differences in means and Z-statistics from two-sample Wilcoxan tests.
Multivariate regression results
The prime test regarding H1 and H2 is around the employed logit model. H1 and H2 are tested under multivariate regression and the results are presented in Table 6. According to that χ2=436.12, p=0.000, the results are statistically significant. The value of Pseudo R2 is 0.1700, representing an acceptable fit of the model.
Table 6
Multivariate regression results
Var. Predicted sign Coefficients
EVOL (-) -155.162*** (-9.53) SIZE (+) 0.248*** (9.00) LEV (+) 5.461*** (14.21) ROA (+) -71.955*** (-9.59) CFVOL (-) -22.825*** (-4.35) PB (+) 0.076 (1.62) o.YR_Dum2 0.000 (.) YR_Dum3 0.688* (1.90) YR_Dum4 0.462 (1.30) YR_Dum5 0.135 (0.42) YR_Dum6 0.021 (0.07) YR_Dum7 -0.050 (-0.16) YR_Dum8 -0.205 (-0.66) YR_Dum9 -0.132 (-0.44) YR_Dum10 -0.140 (-0.49) YR_Dum11 -0.037 (-0.13) YR_Dum12 -0.124 28
(-0.44) YR_Dum13 -0.124 (-0.45) YR_Dum14 -0.145 (-0.53) YR_Dum15 -0.206 (-0.75) o.YR_Dum16 0.000 (.) Constant -6.247*** (-11.14) No. of observations 4339 Pseudo R^2 0.1700 χ2 436.12 0.000
Notes: Table 6 presents regression results of the following logit model, where L equals 1 if a bank writes comment letter during the IASB due process (represent lobbying activity) and 0 otherwise:
L = β0 + β1EVOL + β2SIZE + β3LEV + β4ROA + β5CFVOL + β6P/B + β7YR_Dum2 + β8YR_Dum3 +
β9YR_Dum4 + β10YR_Dum5 + β11YR_Dum6 + β12YR_Dum7 + β13YR_Dum8 + β14YR_Dum9 +
β15YR_Dum10 + β16YR_Dum11 + β17YR_Dum12 + β18YR_Dum13 + β19YR_Dum14 + β20YR_Dum15 +
β21YR_Dum16 + ε.
Variables are clarified in the methodology section. Coefficients and z values are shown in the third column. ***, ** mean that coefficients are at significant level of 1%, 5% (one-tailed). YR_Dum2 and YR_Dum16 omitted because of collinearity.
H1 is supported by the EVOL coefficient -155.162, significant at 1% level. It indicates that lobbying entities are of lower earnings volatility and entities of higher earnings volatility lobby less. However, both the prediction and results are different with prior empirical research. Ang (2000) predicts positive sign regarding the variable income volatility and achieves positive but less significant correlation with lobbying activities. Kosi (2014) also predicts positive sign, while his result is negative but insignificant coefficient. Such distinguishes may be due to different operating industry of samples and different time period focused by researchers.
As to H2 saying that lobbying entities are with higher leverage ratio (LEV), the result also corresponds to such prediction, given the coefficient of 5.461 and significant at 1% level.
The finding is consistent with a U.S. study, Ndubizu et al. (1993), achieving that leverage structure and earnings volatility are all determinants driving lobbying behaviours. On the contrary, European literatures achieve disunion results and show no significance (Jorissen 2012; Kosi 2014). One reason could be that although lobbying firms have higher leverage, it is not a vital determinant to the firm’s lobbying activity. While result of this study verifies the positive relationship predicted, indicating that firms of higher operating risk tend to lobby more.
Regarding control variables, SIZE is significant at 1% level with positive sign. Sutton (1984) notes that larger firms are more motivated to lobby because they usually obtain higher benefits from lobbying and such benefits are adequate to cover lobbying costs as they own more wealth than small firms. Watts and Zimmerman (1978) mention that larger firms incur more political costs, thus size captures the extent of political costs in lobbying activities. Their arguments could serve as explanations for the positive sign obtained in this research. ROA is also significant at 1% level. However, the relationship is negative, contrary to the predict sign. As ROA represents profitability (Jorissen, 2012), one possible explanation could be that firms of higher profitability depend less on external stakeholders and are more capable to bear changes in regulation environment. Accordingly, they lobby less than those of less profitability.
CFVOL is significant at 1% level and shows the same sign compared to EVOL. The absolute value of CFVOL coefficient is smaller than that of EVOL, acting as potential evidence that cash flows volatility contributes to earnings volatility. As for PB, it shows no significance under the multivariate regression. This is consistent with the insignificant correlation coefficient obtained in Table 3 (Panel D). Therefore, the PB capturing growing opportunities of an entity is not a determinant and not very relevant in driving a firm’s lobbying behaviour. Looking at year dummies, the coefficients are all insignificant, indicating the representativeness of the time period employed in the research. The numbers are positive at first several years while started to turn to negative ones in year 2004 (YR_dum7). I guess one reason is that after IFRS implementation becomes mandatory, banks consider weakened influence they could have on the content of accounting standards, thus the year dummy coefficients remain negative since then.
4.2 Robustness test
To confirm the robustness of results attained in this study, I perform robustness test by employing a constrained sample pool. I only focus on banks in the same region with that of lobbying banks to enhance the comparability, as differences in various aspects exist among countries. The modified sample pool is limited to 1,331 observations from 2,022 figures after deleting missing values within the same time period of 1998-2013. Descriptive statistics and testing results are reported in Table 7 and Table 8 respectively.
Table 7 Descriptive Statistics
Panel A: Distributional Statistics
Variable N Mean Std. Dev. Q1 Median Q3
LETTER 2,022 0.257 0.437 0.000 0.000 0.000 EVOL 1,531 0.007 0.007 0.003 0.005 0.008 SIZE 1,935 16.084 1.900 14.606 16.009 17.556 LEV 2,017 0.253 0.145 0.151 0.235 0.335 ROA 2,018 0.008 0.011 0.004 0.007 0.011 CFVOL 1,437 0.014 0.013 0.005 0.010 0.018 PB 1,933 1.774 1.095 1.024 1.557 2.285
Panel B: Correlation coefficients
Var. LETTER EVOL SIZE LEV ROA CFVOL PB
1 LETTER 1 -0.2785 0.4392 0.1551 -0.2882 -0.0358 -0.0761 2 EVOL -0.2118 1 -0.0799 -0.0157 0.0192 0.1953 -0.0895 3 SIZE 0.4386 -0.1625 1 0.1217 0.0509* 0.3192 0.1442 4 LEV 0.1157 0.0352 -0.0018 1 -0.2078 0.2529 -0.0622** 5 ROA -0.1607 0.0605** -0.0580* 0.1103 1 0.0832 0.6019 6 CFVOL -0.0662** 0.1663 0.2163 0.2083 0.2293 1 0.0288 7 PB -0.0758 -0.0269 0.1359 -0.0128 0.4523 0.0864 1
Notes: Panel A reports distributional statistics of the limited sample. Definitions of variables are in the methodology section. In Panel B above the diagonal are Spearman correlation coefficients and below are the diagonal are Pearson’s correlations. Boldfaced values are significant at 1% level. **, * refers to significant level of 5%, 10%.
Table 8
Frequency Statistics
Country Freq. Percent Cum. AUSTRALIA 112 5.54 5.54 AUSTRIA 107 5.29 10.83 CHINA 95 4.70 15.53 FRANCE 177 8.75 24.28 GERMANY 139 6.87 31.16 SPAIN 129 6.38 37.54 SWITZERLAND 306 15.13 52.67 UNITED KINGDOM 192 9.50 62.17 UNITED STATES 765 37.83 100.00 Total 2,022 100.00
Notes: Table 8 shows frequencies of countries that have banks response in the IASB due process. The time period is 1998-2013. Table 9 Univariate results Section A Non-lobbying observations EVOL: n = 1098 LEV: n = 1497 Section B Lobbying observations EVOL: n = 433 LEV: n = 520 Section C Test of H0: A = B Mean Median Std.
Dev. Mean Median
Std. Dev. t-statistics (p-value) z-statistics (p-value) EVOL 0.008 0.006 0.008 0.005 0.004 0.004 8.261 10.038 0.000 0.000 LEV 0.247 0.228 0.151 0.271 0.258 0.126 -3.288 -4.661 0.001 0.000
Notes: T-statistics are from tests on differences in means and Z-statistics from two-sample Wilcoxan tests.
Table 10
Multivariate testing results
Var. Predicted sign Coefficients
EVOL (-) -256.402*** (-10.11) SIZE (+) 1.086*** (14.75) LEV (+) 2.792*** 32
(3.48) ROA (+) -167.942*** (-10.30) CFVOL (-) -56.143*** (-6.31) PB (+) -0.470*** (-4.00) o.YR_Dum2 0.000 (.) YR_Dum3 1.739*** (3.29) YR_Dum4 1.219** (2.39) YR_Dum5 1.359*** -2.85 YR_Dum6 1.190** (2.49) YR_Dum7 0.722 (1.58) YR_Dum8 0.585 (1.27) YR_Dum9 0.700 (1.54) YR_Dum10 0.409 (0.97) YR_Dum11 0.535 (1.33) YR_Dum12 0.570 (1.41) YR_Dum13 0.586 (1.46) YR_Dum14 0.236 (0.60) YR_Dum15 -0.097 (-0.24) o.YR_Dum16 0.000 (.) Constant -16.650*** (-13.16) No. of observations 1331 Pseudo R^2 0.3998 χ2 632.04 33
0.000
Note: Table 10 reports results of the following regression model:
L = β0 + β1EVOL + β2SIZE + β3LEV + β4ROA + β5CFVOL + β6P/B + β7YR_Dum2 + β8YR_Dum3 +
β9YR_Dum4 + β10YR_Dum5 + β11YR_Dum6 + β12YR_Dum7 + β13YR_Dum8 + β14YR_Dum9 +
β15YR_Dum10 + β16YR_Dum11 + β17YR_Dum12 + β18YR_Dum13 + β19YR_Dum14 + β20YR_Dum15 +
β21YR_Dum16 + ε.
L equals 1 if a bank writes comment letter under current stage of IFRS 9 project (represent lobbying activity) and 0 otherwise. Variables are clarified in the methodology section. Coefficients and z values are shown in the third column. ***, ** refer to coefficients at 1%, 5% significant level (one-tailed). YR_Dum2 and YR_Dum16 omitted because of collinearity.
Differences exist between correlation coefficients of original (Table 3, Panel D) and constrained sample (Table 7, Panel B). Looking at Pearson’s coefficients above the diagonal, PB shows negative and significant value, ROA becomes positive and significant, LEV becomes positive. Coefficients between EVOL and LEV, EVOL and ROA turn to be insignificant. Significance shown in Table 9 verifies hypothesises in univariate level.
For Table 10 reporting multivariate regression results, the 0.000 p value shows the results are statistically significant. The Pseudo R2 is 0.3998, meaning good fit of the model. The EVOL and LEV coefficients show the same relationship with original results, confirming both H1 and H2. The absolute value of EVOL is larger than that attained in the original sample, enhancing the predicted influence of earnings volatility on lobbying activity. LEV is significant while decreases from 5.461 to 2.792, reflecting a weaker driving effect compared to sample selected all over the world. Considering other variables, the absolute value of SIZE, ROA are larger than that in Table 6. This indicates stronger correlation and more lobbying activities would be conducted if these variables change same scale within these countries compared to the wider sample scope. However, PB coefficient turns to be negative and significant, contrary to 0.076 in the initial results. This may be caused by decreased countries diversity and decreased noising. Growing opportunities captured by PB serve as a considerable characteristic and affect a firm’s lobbying decision. One explanation of the negative sign is that banks with more grow opportunities are less constrained by the revision of accounting standards and thus lobby less.
Year dummies are all positive expect -0.097 in 2012, while they show a decreasing trend on the whole. Overall, the robustness test confirms the attained results and supports the two hypothesises of this study.
5. Conclusion
This study investigates characteristics of lobbying and non-lobbying banks towards IFRS 9 and focuses on lobbyists responding to Classification and Measurement: Limited Amendments to IFRS 9 (Proposed amendments to IFRS 9 (2010)) issued by the IASB under the IFRS 9 project's current stage. As the potential effects of revised standards on a firm's earnings volatility, the study examines how earnings volatility characteristic varies among these lobbying banks and non-lobbying banks. A binary model is established and negative relationship between earnings volatility and lobbying activities is achieved.
I find that banks conducting lobby activities towards the IFRS 9 newest exposure draft show lower earnings volatility than those not lobbying. The first hypothesis is verified. One probable explanation for such relationship is that the new amendments of IFRS 9 would affect a firm's earnings volatility. Banks with earning figures maintain stable prior to the issue of revised standards tend to face more changes in their earnings streams. As stable earnings streams are valued by investors and other externals, these banks will suffer greater loss and bear more doubts compared to those with more volatile earnings previously because their earnings streams change relatively less. Looking at modifications of the new ED, the SPPI test (solely payment of principal and interests test) would contributes directly to the changes in a bank’s earnings streams. If a financial instrument failed the test at first step in judging its classification and measuring, it would be measured at FVTPL. There are financial instruments measured at amortised under IAS 39 but would fail such test under the new IFRS 9, leading to less stable earnings streams as fair values are adopted to more of the financial instruments. Meanwhile, I find that banks with higher leverage ratio tend to lobby more. I explain this positive sign as that high leverage ratio implicates high risks and higher borrowing costs, banks are reluctant to any changes made in accounting standards that would additionally increase their leverage ratio. Furthermore, size and ROA also show significant correlations with a bank's lobbying activity. The fixed-year effects show a downward trend of lobbying. I attribute this to a better standardized operating environment and various emerging regulations a bank should obey in the banking industry. Or the voices from larger and famous banks are thought to be representative by some smaller and less famous ones.
On the whole, the study contributes to lobbying and standard-setting literature. As banking industry operates with various and complicated financial instruments, considering characteristics of banks and their lobbying motives would promote the development of IFRS 9 and help to achieve higher quality of the standards.
However, limitations exist in this study. First, variables used in the regression model are not perfect ones. Some other factors that may affect lobbying behaviours are not included, for instance, whether the bank adopts IFRS or the US GAAP, whether it has lobbied before current stage, and the type of a bank. Second, during the time period of data employed the 2008 financial crisis happened, heavily volatile earnings streams appeared in such crisis. Therefore, some of the data may be noising for hypothesis test. Third, as I investigate responding banks merely in current stage of IFRS 9 project, the amount of lobbying sample may be not rich enough compared to the non-lobbying sample.
Future researches could complement this study through establishing a more complete regression model containing other influencing factors, investigating lobbying activities throughout the whole IFRS 9 projects, looking into the effects of some other characteristics on lobbying activities, or using a wider range of sample instead of mere banks. In addition, the IASB is incorporating amendments on impairment and classification and measurement, a final version of IFRS 9 is expected to come out soon. To what extent the IASB responses to the issues raised in comment letters submitted and make relative changes in standards is worthwhile studying.
Acknowledgements
I thank Dr. Sanjay Bissessur for his helpful guidance and suggestions, the work of Dr.Ir.Sander van Triest, the library sources of University of Amsterdam, and the IFRS site with all the comment letters available.
