The value relevance of IFRS 16 on the Swedish market

(1)

Master’s Thesis 30 credits

Department of Business Studies Uppsala University

Spring Semester of 2020

Date of Submission: 2020-08-06

The value relevance of IFRS 16 on the Swedish market

Elin Pettersson

Matilda Hansson Brusewitz

Supervisor: Joachim Landström

(2)

2

The value relevance of IFRS 16 on the Swedish market

1Elin Pettersson, ¹Matilda Hansson Brusewitz

1Department of Business Administration, Uppsala University, Uppsala, Sweden

1E-mail: Elin.Pettersson.9058@student.uu.se,

1Matilda.Hanssonbrusewitz.8180@student.uu.se Received: The 6^th of August 2020

Abstract

The new standard IFRS 16 regarding leases was implemented 1st of January 2019 with the aim to improve accounting for leases and to provide more faithful information on the financial statements. We are conducting a value relevance study to observe to contribution of IFRS 16 to more value relevant information for investors and analysts. We are adopting an adjusted version of the valuation framework developed by Ohlson (1995) to be more coherent with the expected changes due to the new standard. We measure changes in value relevance pre and post the implementation of IFRS 16 using 8 quarterly interim reports representing the fiscal years 2018 and 2019. Furthermore, we are adopting an industry adjusted model to measure if value relevance has changed more in lease prevalent industries since industries such as retail, airline and service industries are expected to experience larger changes due to the new standard. We are considering 262 companies listed on the Swedish stock market, NASDAQ Stockholm. Our results suggest that the value relevance has increased in the post period of the implementation, especially in lease heavy industries.

Keywords: IFRS 16, IFRS implementation, lease liabilities, value relevance study, Ohlson model

______________________________________________________________________________________

(3)

3

Content

1. Introduction ...4

2. Background ...6

3. Literature review ...7

3.1 Leases and debt ...7

3.2 Capitalization of leases ...7

3.3 Recognition versus disclosure ...8

3.4 Value relevance ...9

3.5 Hypothesis Development ...10

4. Methodology ...12

4.1 Sample and data ...12

4.2 Research methodology ...13

4.3 Specification tests ...17

4.4 Diagnostic tests for autocorrelation, heteroscedasticity and cross-sectional dependence ...17

5. Empirical results ...18

5.1 Univariate analysis ...18

5.2 Bivariate analysis ...18

5.3 Multivariate analysis ...19

6. Conclusion ...26

References ...28

Appendix ...32

(4)

4

1. Introduction

During the last decades, concerns have been expressed regarding if financial statements have started to lose their relevance for valuation purposes (Francis & Schipper, 1999). Relevance is lost if the information does not have the ability to influence decision making regarding providing resources to the entities (IFRS Foundation, 2018). Accordingly, the International Accounting Standards Board (IASB) are continuously working towards improving accounting information to be relevant and faithfully presented. One step being the development of IFRS 16 Leases.

Under IFRS 16 accounting information is supposed to be more transparent and more faithful regarding the firm’s financial position. With IFRS 16 the previous categorization between operating leases and financial leases prevailing under IAS 17 is removed (IASB, 2016). The removal of the categorization implies that operating leases are capitalized. Consequently, leverage and EBITDA are significantly affected, which is expected to be more distinct in lease prevalent industries (Imhoff et al., 1991; Goodacre, 2003; Mulford &

Gram, 2007; Fülbier et al., 2008; Duke et al., 2009; Singh, 2012; IASB, 2016). Furthermore, the categorization implied a risk of deceptive assessments of the financial health of firms because of the opportunity to use assets that were not required to be recognized on the balance sheet (Imhoff et al., 1991; Fülbier et al., 2008; Duke et al., 2009). As operating leases have been off-balance sheet leases it has also led to beneficial debt-ratios (Petersen et al., 2017). Furthermore, lease financing has been favorable due to the advantage of being cheaper compared to corporate loans and other funding methods.

The IASB (2016) has stated that the absence of information about leases on the balance sheet, and the significance of missing information, has varied by industry and region, which complicates comparability for investors and analysts (IASB, 2016). Hence, IFRS 16 aims to improve comparability between firms using different financing methods and between firms in different industries (Imhoff et al., 1991; IASB, 2016).

Consequently, accounting information regarding leases should become more value relevant after the implementation, as IFRS 16 forces firms to present a more faithful view of their financial position. By examining the value relevance of IFRS 16, establishments can be made on how well particular accounting amounts such as leverage and EBITDA reflect information that is used by investors in valuing the firm’s equity value and assess the relevance and reliability of accounting amounts the IASB aims to fulfill (Barth et al., 2000).

However, previous studies suggest that operating leases were assessed by investors and analysts pre IFRS 16 (Beattie et al., 2000; Giner & Pardo, 2018). Arguing for leases being a substitute to corporate debt (Myers et al., 1976; Deloof & Verschauren, 1999). Thus, independent from different recognition and disclosure requirements, the changes in financial ratios due to capitalization of operating leases should be irrelevant when assessing firms’ financial risk. Suggesting that information provided in the financial statements after the implementation of the standard would not become more value relevant.

On the other hand, a set of studies examining value relevance of information after adopting IFRS as a general framework suggest that some accounting amounts increases value relevance (Karğın, 2013; Kouki, 2018). Additionally, several studies suggest that information becomes more value relevant together with added recognition requirements, highlighting the issue of disclosed information not being sufficient for a proper

(5)

5

analysis, arguing in favor of IFRS 16 increasing value relevance (Davis-Friday et al., 2004; Ahmed et al., 2006;

IASB, 2016). Accordingly, the user expertise and accessibility of additional sources of information matter as assessment of operating leases is expected by expert users, but financial statements are characterized by information deficiency from a non-expert user perspective (IASB, 2016). Hence, the necessity of recognition requirements and the importance of more value relevant information regarding leases is crucial for smaller investors relying on financial statements as the sole source of information. Thus, comprising an argument of IFRS 16 contributing to more value relevant information. Furthermore, some industries should be more sensitive to the adoption of IFRS 16 as they rely more on leasing, raising the question of whether lease prevalent industries potentially have experienced a larger change in value relevance of the information (Mulford & Gram, 2007; Singh, 2012). Hence, the purpose of the thesis is to examine the remaining question of whether IFRS 16 has contributed with more value relevant information for investors and analysts, in general, and in lease heavy industries.

This study complements previous literature by considering the first financial reports including the IFRS 16 implementation. Previous studies have been conducted pre IFRS 16 and been required to capitalize operating leases to investigate the potential value relevance effects. Thus, this study contributes with the actual value relevance effect as we are using the officially reported leasing liabilities. Furthermore, the IASB has stated that information has been significantly varied by region. Therefore, this study contributes with information regarding its impact on the value relevance of accounting information for Swedish listed companies specifically as no similar study has previously been conducted in the Swedish context. Lastly, the study complements previous literature by investigating the impact of value relevance on an industry level.

Our results indicate that IFRS 16 has contributed to more value relevant information for investors and analysts, with slightly stronger empirical results when adopting an industry specific model focusing on lease prevalent industries.

The thesis continues with the following structure. Section II describes the background of IFRS 16. Section III constitutes of the literature review, including our hypothesis development. Section IV describes our methodology and section V provides empirical results and discussion. Lastly, section VI constitutes of the conclusion.

(6)

6

2. Background

According to the conceptual framework of IFRS, the objective of financial reporting is “to provide financial information that is useful to users in making decisions relating to providing resources to the entity” (IFRS Foundation, 2018). The conceptual framework states out qualitative characteristics that stipulate the usefulness of financial information which is met when information is relevant and provides a faithful representation.

Information is considered relevant when it has the capability of making a difference in the decisions made by users of the financial statements. Users of financial statements in the conceptual framework are defined as both existing and potential investors, lenders and other creditors who must rely on general purpose financial reports for much of the financial information they need (IFRS Foundation, 2018). The primary users in this study are smaller non-expert investors, as the only source of information they have access to is the public information in financial statements in comparison to managers, institutional investors and creditors who have access to other sources of information.

From the first of January 2019, IFRS 16 has been effective as the new standard managing lease accounting. The previous standard IAS 17 was according to IASB (2016) focusing on dividing leases into either an operating lease or a finance lease. The determination was based on the economical similarity of purchasing the underlying asset. Hence, depending on what kind of lease, it was accounted for differently. A finance lease was required to be recognized on the balance sheet whereas operating leases only had disclosure requirements (IASB, 2016). The difference between recognition and disclosure is grounded in the incorporation of information in the financial statements. Recognition implies an item is being recognized with both words and numbers on the financial statement whereas disclosure implies information is given by the notes about the item (FASB, 1984). Following IFRS 16, all leases have recognition requirements. In regard to the conceptual framework, the previous IAS 17 was in part contradicting the main objective of financial reporting as the categorization of leases implied the possibility of misleading information was presented on the financial statements, hence not a faithful presentation of the firm’s financial position. Furthermore, the IASB (2016) pointed out that comparisons between firms with different financing methods, a firm that has chosen to buy their assets and a firm that has chosen to lease their asset, was not possible to perform accurately.

Technically, the switch from IAS 17 to IFRS 16 implies all leases are required to be recognized on the balance sheet instead of as a lease expense in the income statement (IASB, 2016). The visible changes on the balance sheet are an increase in lease assets. The increase is due to the additional leases being capitalized and accounted for as the present value of the lease payment and subsequently recognized as a lease asset or possibly as a part of property, plant and equipment. Another visible change is an increase in financial liabilities which compose of eventual future lease payments (IASB, 2016). The most significant changes due to IFRS 16 are connected to the balance sheet, however, the new standard also implies changes concerning the income statement. The operating lease expense is being removed as a result of IFRS 16 and instead, there is a depreciation of the leased asset and an interest expense deriving from the lease liability that burdens the income statement. Therefore, the visible changes in the income statement are an increase in EBITDA and an increase in operating profit (IASB, 2016).

(7)

7

3. Literature review 3.1 Leases and debt

Academic financial accounting theory suggests that lease liabilities are similar to long-term debt due to their non-cancellable agreements (Schroeder et al., 2011). Traditional finance theory suggests that corporate debt and leases are substitutes, because both are fixed and contractual obligations that reduce the firm’s debt capacity (Myers et al., 1976; Deloof & Verschauren, 1999). Previous studies support that leases displace debt, as an increase in corporate debt is significantly associated with a decrease in lease financing (Marston & Harris, 1988;

Lewis & Schallheim, 1992; Krishnan & Moyer, 1994; Adedeji & Stapleton, 1996). Moreover, Bayless and Diltz (1986) suggest that bank officer’s willingness to lend to a firm is reduced when the firm takes on more lease obligations, which suggests that they are too similar in their characteristics.

Contradicting, some studies have found evidence that debt financing and lease financing are complements and not substitutes. Ang and Peterson (1984) found that lessee firms use more long-term corporate debt than non-leasing firms, which suggests that leases and debt are complements. The puzzle between these conflicting results was further investigated by Marston and Harris (1988). Their results support that leases and debt are substitutes, by comparing changes rather than levels, together with recognition of debt capacities between different corporations (Marston & Harris, 1988). A more recent study made by Giner and Pardo (2018) has found similar results suggesting that investors do not value off-balance sheet leases differently to corporate debt.

Thus, the leasing puzzle is central in how the new accounting amounts provided on the financial statements due to IFRS 16 are interpreted by investors and analysts. The view on leasing liabilities in comparison to corporate debt influences the magnitude of the expected effects due to capitalization of operating leases when analyzing lease information. Consequently, influencing if the accounting information becomes more value relevant or not.

3.2 Capitalization of leases

Capitalization of operating leases affects vital accounting amounts and related financial ratios which could influence how firms are valued. Additionally, different capitalization methods have been used among investors in order to estimate lease liabilities (IASB, 2016). Thereby, possibly affecting value relevance of financial information for investors and analysts. Accordingly, the IASB (2016) expected an increase in leverage and an increase in EBITDA due to capitalization of operating leases under the IFRS 16 regulations. Moreover, ex ante- studies focusing on capitalization of leases have a coherent view concerning significant changes in financial ratios. Imhoff et al. (1991) investigated the effects on the ratio debt-to-equity among other ratios of a total of 14 firms paired up to capture both inter- and intra-industry differences. The methodology of Imhoff et al. (1991) on capitalizing leases is especially interesting because it is used for a major part of the literature attempting to investigate the effects of capitalizing leases, hence, potential effects of IFRS 16. Their findings suggest that if capitalization of operating leases would occur, the leverage ratio would show an expected average increase of 191 percent for lease heavy firms and 47 percent for less lease heavy firms (Imhoff et al. 1991). Furthermore, Fülbier et al. (2008) showed major changes in terms of the relationship between debt and equity when examining German firms but with moderate results compared with prior studies. Duke et al. (2009) provide results that are

(8)

8

in accordance with Imhoff et al. (1991) showing a damaging effect on the debt-to-equity and debt-to-asset ratios, showing significant increases. Moreover, Imhoff et al. (1997) illustrate that capitalization of leases will impact net income in terms of a decrease. Morales-Díaz and Zamora-Ramírez (2018) conducted one of the last ex-ante studies but used an adapted capitalization model more in line with IFRS 16 as the final issuance of IFRS 16 had been released. The results indicate an increase in leverage measured as both debt-to-equity and debt-to-assets and an increase in EBITDA. However, their results point significantly to the differences between industries.

3.2.1 Capitalization of leases in lease prevalent industries

The effect analysis by IASB (2016) on the new leasing standard states that IFRS 16 affects industries relying more on leases, such as airline, retail, and travel and leisure. In line with the IASB, previous research provides evidence for some industries being more lease heavy than others. Krishnan and Moyer (1994) confirm that transportation is a lease prevalent industry. Furthermore, it is stated by Adams and Hardwick (1998) that

‘services and utilities’ is a lease intense industry based on a sample of UK firms. Gosman and Hanson (2000) further state that airlines and the retail industry, including restaurants, are industries known to be lease heavy.

Furthermore, studies have investigated the effects due to capitalization of operating leases in lease heavy industries. Mulford and Gram (2007) focused their study on the retail industry and found a median increase in EBITDA representing 22.5 percent and a median increase representing 26 percent when investigating debt-to- equity for retail companies. Similarly, the study by Singh (2012) is delimited to an industry-specific context when evaluating the influence of capitalizing operating leases on financial ratios. More precisely the study focused on retail and restaurants showing significant results. In terms of the debt to equity ratio Singh (2012) suggests an increase of more than three times, from 0.30 to 1.38 equalizing an increase of 354 percent.

Moreover, according to Morales-Díaz and Zamora-Ramírez (2018), the sectors most affected by lease capitalization are retail, transportation, hotels, software and service sectors. Giner and Pardo (2018) suggest that off-balance sheet leases are significant in relation to market value in the retail sector but not in other sectors.

3.3 Recognition versus disclosure

As IFRS 16 implies a change from previous disclosure requirements to current recognition requirements, the recognition versus disclosure debate is a central aspect of the value relevance of the accounting information regarding leases. The recognition versus disclosure debate has increased in significance and focus on investigating if disclosed information is perceived equally to recognized information on financial statements (Beattie et al., 2000). The market efficiency theory assumes recognition and disclosure to be equal, however, it assumes that investors analysts have not been misled by operating leases and how these were disclosed in financial statements (Beattie et al., 2000; Goodacre, 2003).

Beattie et al. (2000) provide evidence implying that investors and analysts in the UK are incorporating operating leases in their risk assessment of the firms, suggesting that operating leases were being assessed by investors and analysts even without IFRS 16 regulations arguing for disclosed information being equal to recognized information. Coherently, a more recent study by EFRAG (2017) shows that the market seems to capture the leasing liability even though there is a different treatment for accounting purposes. Moreover,

(9)

9

Altamuro et al. (2014) find that credit markets and credit rating agencies incorporate leasing information into their credit assessments. Contradicting evidence concludes that the accounting method for operating leases has a significant role as the study by Breton and Taffler (1995) provide findings indicating that adjustments for operating leases are not made in the assessment of financial ratios. Further research finds evidence that the market participants price operating lease commitments. Sengupta and Wang (2011) find that recognized liabilities have a stronger effect than disclosed ones, whereas Lim et al. (2017) finds no evidence. Moreover, it has been pointed out by the Capital Markets Advisory Committee that the previous requirements for disclosure regarding operating leases may be enough for expert investors, but it is not enough disclosure for non-expert investors. Another problem raised is the different methods used by investors and analysts to adjust financial statements when considering operating leases (IASB, 2016).

3.4 Value relevance

The major part of the value relevance studies uses the valuation framework by Ohlson (1995) when attempting to investigate the implementation of new accounting standards in terms of recognition versus disclosure. Ahmed et al. (2006) conducted a value relevance study on disclosed derivatives prior to SFAS No. 133 which post the implementation required recognition. Their findings suggest that with only disclosure requirements the derivatives are not value relevant but are value relevant when required to be recognized. Furthermore, Davis- Friday et al. (2004) investigated if the market considered disclosed information less reliable compared with recognized information by conducting a value relevance study on postretirement benefits other than pensions.

Their results indicate that disclosed information is viewed as less reliable than recognized information, hence, showing consistent results with Ahmed et al. (2006). Furthermore, Giner and Pardo (2018) used the Ohlson model to investigate the value relevance of IFRS 16 before the implementation in a Spanish context. The utilized model includes capitalized operating lease liabilities, recognized debt and net income as control variables and aims to identify how the Spanish market values operating lease liabilities. Their results show that the market views operating leases as liabilities and suggests that investors perceive all leased assets as property rights.

Additionally, Giner and Pardo (2018) compared retail companies and non-retail companies specifically. The results show that the operating lease liability coefficient was negative and significant for retail companies, but not significant for non-retail companies. Hence, the results of Giner and Pardo (2018) indicate that value does not differ between recognized debts and off-balance sheet liabilities. Suggesting that the recognized information is not more value relevant than disclosed information.

Moreover, another set of studies examining value relevance when adopting the general IFRS framework.

Martínez et al. (2014) examine the mandatory adoption of IFRS on Spanish firms and the related market reactions. They conduct a value relevance analysis comparing accounting numbers under IFRS and the local GAAP with results suggesting that overall, accounting numbers under IFRS is not more value relevant compared with the local GAAP. Karğın (2013) used the Ohlson model to measure the market value of equity and financial reporting variables of IFRS standards on Turkish listed firms. Karğın (2013) compares book values per share and earnings per share pre and post the implementation. The results show that value relevance of

(10)

10

accounting information increases significantly in the post IFRS implementation period for the book value per share, while the value relevance of earnings per share has decreased. Kouki (2018) examined if the relevance of equity book value, earnings, and changes in earnings improved in Germany, France and Belgium by comparing IFRS-firm and non-IFRS-firms before and after the implementation of IFRS. The findings suggest that the implementation of IFRS increases the value relevance of equity book value and earnings. However, the assumption of voluntary IFRS adoption in the pre-period of the mandatory implementation should increase value relevance is not confirmed. Mirza et al. (2019) examined value relevance of the IFRS adoption in Malaysia using a developed version of the Ohlson model with firm-specific control variables, adopting book value, cash flow, total assets, leverage in relation to market value. Mirza et al. (2019) find that both earnings and book value are significant value relevant for investment decision making and that financial reporting is generally valuable.

A further finding is that cash flows provides more relevant information to Malaysian investors, in comparison to the statement of financial position. Odoemelam et al. (2019) examined earnings incremental value relevance as an effect of adopting IFRS in Nigeria by using an adjusted Ohlson model including dummy variables. Their findings suggest that earnings value relevance increases with the adoption of IFRS, but the book value of equity value relevance does not.

3.5 Hypothesis Development

If leases have been considered as substitutes to corporate debt as suggested by Myers et al. (1976) and Deloof and Verschauren (1999), the change in financial ratios would not affect the assessment of the firm’s financial position and would not imply more value relevant information for investors and analysts. However, it implies that the previous discloser requirements have been sufficient to value operating lease liabilities. Though, IASB (2016) states that there has been a significant absence of information regarding lease financing in the financial statements. Therefore, the Board has expressed that the new leasing standard is expected to result in more faithful information being presented on firms' financial statements, resulting in more value relevant information.

Thus, due to the new recognition requirements of leases, the opportunity has occurred to test whether the new accounting standard has made an impact on value relevance of the accounting information provided in the financial statements. Previous literature indicates that adoption of IFRS as a general framework have, in part, increased value relevance (Karğın, 2013; Mirza et al., 2019). Contradicting that assumption, another set of studies suggest that previous operating leases with disclosure requirements have been incorporated when assessing the firm’s financial risk. Arguably, accounting information post the IFRS 16 implementation should not be more value relevant for investors and analysts (Beattie et al. 2000; Altamuro et al. 2014). On the other hand, previous studies indicate that recognized information is more value relevant than disclosed information arguing favorably for the contribution of more value relevant information due to IFRS 16 (Davis-Friday et al., 2004; Ahmed et al., 2006). Moreover, as it has been pointed out by Capital Markets Advisory Committee, disclosed information is most likely not sufficient for non-expert investors. Therefore, the expectation is that IFRS 16 increases value relevance as smaller outside investors are the ones most dependent on the information provided in the financial statements. Based on previous studies suggesting that information becomes more value

(11)

11

relevant in conjunction with adoption of IFRS and in accordance with the expectations of the IASB, we expect IFRS 16 to increase value relevance for investors and analysts.

Hypothesis I: IFRS 16 increases value relevance

IASB (2016) states that the missing information about lease financing in the financial statements has been varied between industries and regions which complicates comparability. Coherently, studies within leasing are providing a consistent view regarding some industries being more lease heavy than others. Industries such as transportation, airlines, retail, hotels, and restaurants are identified as lease intense industries (Krishnan &

Moyer, 1994; Gosman & Hanson, 2000; Mulford & Gram, 2007; Singh, 2012; Morales-Díaz & Zamora- Ramírez, 2018). Consequently, expectations are that changes regarding the financial information after the implementation is more drastic in lease intense industries. Hence, as IFRS 16 aims to provide more transparent and valuable information about leases for investors and analysts, we expect the accounting information to be more value relevant in lease heavy industries.

Hypothesis II: IFRS 16 increases value relevance depending on industry

(12)

12

4. Methodology 4.1 Sample and data

This study is conducted on the Swedish stock market. Hence, the sample collection is restricted to the Nasdaq Stockholm exchange. Nasdaq Stockholm requires IFRS standards as the stand-alone official financial reporting standard setter (Finans Inspektionen, 2019). Therefore, unregulated stock exchanges are excluded as they are not always fully restricted to the IFRS standards but are also allowed to adopt the Swedish local GAAP, K3.

The database used for both sample collection and financial data collection is Bloomberg L.P. The sample collection is manually designed within Bloomberg’s database and includes all equities with the currency “SEK”, the exchange “Sweden Stockholm”, with the primary exchange name “Stockholm Nasdaq”, and with the equity type “Common stock”. The sample consists of a total number of 338 firms. Firms with a broken financial year are excluded because they do not report the IFRS 16 for the first quarter of 2019. Moreover, some collected data has reported insufficient information to Bloomberg and is identified as “Not available”, which prevented all required data from being properly collected. The number of firms with broken financial year and firms with data deficiencies, amounts to 23. Furthermore, 53 companies have not reported the specific financial item EBITDA for all periods to Bloomberg L.P and have been excluded from the sample. The method of handpicking the inaccessible data manually is excluded to maintain a standardized collection method. Thus, the final number of firms included in the study amounts to 262. Total number of excluded securities are 76 and represents ~22.5 percent of the original sample. Dual class shares are included to observe the full market effects. The financial data collected for each firm is based on Bloomberg’s Fundamentals methodology.

The time period of this study is divided into two parts, the pre IFRS 16 period and the post IFRS 16 period. The pre IFRS 16 period represents the full fiscal year of 2018 and the post IFRS 16 period represents the full fiscal year of 2019. Thus, panel data of eight quarterly interim reports is used for the accounting information. Quarterly reports are chosen with the purpose of increasing the sample size since the study is restricted to a two-year window. Moreover, quarterly reports give a more detailed overview of the financial development during the full fiscal years. However, quarterly reports result in possible seasonal effects in the data set, which is adjusted with quarterly time dummies in the models. The choice of financial data and accounting information variables collected for our models is based on the adjusted model by Mirza et al. (2019).

The dependent MVPS variable corresponds to the closing share price of the firm and is collected quarterly over the fiscal years but collected at three months following the financial quarter-end date. We choose to collect MVPS with a three-month lag to ensure the available information is reflected in the share price since Nasdaq Stockholm release the new accounting information to the public with a one to three-month lag to the book end- date. However, the choice of collecting the share price with a three-month lag instead of four-month lag as made by Mirza et al. (2019) is due to the opportunity to reduce the noise emerging when collecting accounting data with a time lag. Hence, we adjust the collection date to three months after the book end date, whereas each regression will be paired with the market value of equity that has been exposed to the new public financial statement information.

(13)

13 4.2 Research methodology

Value relevance is defined as “the ability of financial statement information to capture and summarize information that determines the firm’s value” (Beisland, 2009). Hence, the main objective is to observe the relationship between market values of equity and accounting variables (Beisland, 2009). Francis and Schipper (1999) states that value relevance can be can determined based on statistical associations between financial information and stock prices or returns. Thus, value relevance can be measured as the ability of financial statement information to capture and summarize information that affects share values which can be conducted using a valuation framework (Francis & Schipper, 1999). According to Beisland (2009), by adopting the interpretation of value relevance by Francis and Schipper (1999), value relevance answer if accounting information has the ability to explain variations in stock prices but does not answer how the information is used (Beisland, 2009).

The valuation framework by Ohlson (1995) is commonly used in value relevance research (Barth et al., 2000). In the original Ohlson model, value is dependent on accounting information, with market value per share as the dependent variable, and earnings, book value and dividends as independent variables (Ohlson, 1995).

The R² is used as the key metric to measure value relevance as it explains how much variance in market value per share is explained by the accounting information in question (Beisland, 2009). Since the Ohlson model was published, studies have contributed with redefinitions of the original model within valuation research, to examine the relationship between market value and variables of accounting information.

Moreover, Beisland (2009) highlights the importance of dissociate relative and incremental value relevance. The difference is defined by Biddle et al. (1995) stating that “incremental comparisons ask whether one accounting measure provides information content beyond that provided by another, and apply when one measure is viewed as given and an assessment is desired regarding the incremental contribution of another (e.g., a supplemental disclosure). Relative comparisons ask which measure has greater information content and apply when making mutually exclusive choices among alternatives”.

We are adopting the interpretation described by Francis and Schipper (1999) using the valuation framework by Ohlson (1995) measuring value relevance using the adjusted R² as a key metric to examine the relative value relevance. We use the adjusted R² instead of R² because our models are multivariate OLS regressions. Hence, an increased adjusted R² value implies increased relative value relevance. Furthermore, we are examining the incremental value relevance by adding an IFRS 16 dummy to our regressions, forcing our independent variables to interact with the IFRS 16 implementation.

4.2.1 Models

Model (1) is a developed model following the adjusted Ohlson model utilized by Mirza et al. (2019) but designed specifically for the purpose of this study. Model (1) is a multivariate OLS regression, including firm- specific control variables, consisting of accounting information items related to IFRS 16. Coherent with the original model by Ohlson (1995) and the adjusted model by Mirza et al. (2019) we are choosing market value per share (MVPS) as our dependent variable which is a proxy for firm value. Coherently with Ohlson (1995)

(14)

14

and Mirza et al. (2019) we choose book value per share (BVPS) as our first independent variable. Moreover, in accordance with the guidelines and supporting materials by IASB (2016), the EBITDA is predicted to change significantly, whereas earnings and cash flow are not (Mulford & Gram, 2007; IASB, 2016). Hence, Model (1) replaces the variable earnings per share used by Ohlson (1995) and by Mirza et al. (2019) with earnings before interest, tax, depreciation and amortization per share (EBITDA), to represent the income statement. In coherence with Mirza et al. (2019) and Odoemelam (2018) we are choosing the natural logarithm of total assets (lnSIZE) and leverage (LEV) calculated as the ratio total debt to total assets, as independent variables, as both are forecasted to increase as a result of IFRS 16 (IASB, 2016). Furthermore, LEV captures both the new recognized lease liabilities, hence the added debt due to the new standard, and the new recognized right-to-use assets, which both are items recognized in the post implementation period. Hence, Model (1) investigates the association between MVPS and our independent variables. Our primary variables are EBITDA, lnSIZE and LEV. Whereas BVPS is chosen as a control variable since it is an essential variable in the original Ohlson model and is included in all adjusted models used in value relevance research. However, book value of equity is not forecasted by previous studies and the IASB (2016) to change significantly due to IFRS 16, therefore, it is not chosen as one of our primary variables. Furthermore, Mirza et al. (2019) are using cash flow of operations and book-to-market ratio in their model which are excluded from our model since the variables are not expected to be influenced due to IFRS 16. Coherently with Ohlson (1995), we are observing the adjusted R² metric closely over time, following the pre-implementation period and the post-implementation period. The adjusted R² metric measures the relative value relevance effect, thereby, if adjusted R² has increased in the post-period, hypothesis I is true.

Model (1) Relative value relevance model

Pre IFRS 16-implementation period

MVPSit = αi + β1BVPSit + β2EBITDAit + β3lnSIZEit + β4LEVit + β5DQ1 + β6DQ2 + β7DQ3 + εi Post IFRS 16-implementation period

MVPSit = αi + β1BVPSit + β2EBITDAit + β3lnSIZEit + β4LEVit + β5DQ1 + β6DQ2 + β7DQ3 + εi

Dependent Abbreviation Measurement and explanation Expected sign

Market value per share MVPSit Measured by the market value of the firm’s share price, at end of three months following the book end date of the financial quarter- end

+

Independent Abbreviation Measurement and explanation Expected sign Book value per share BVPSit Book value of equity per share at the book

end date for the financial quarter-end, at time t and firm i

+

Earnings before interest, taxes, depreciation and amortization, per share

EBITDAi Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares, at the book end date for the financial quarter-end, at time t and firm i

+

Size lnSIZEit Natural log of total assets, at the book end

date for the financial quarter-end, at time t and firm i

+

Leverage LEVit The ratio of total debt to total assets, at the

book end date for the financial quarter-end, at time t and firm i

-

Quarterly dummies DQ1, DQ2, DQ3 Quarterly dummies to adjust for seasonal

effects +

(15)

15

Model (2) is an extension of Model (1), following the utilized model by Odoemelam et al. (2019), who adjust the original Ohlson model by adding a dummy variable to the multivariate OLS regression.

By adding the dummy variables, the model allows to measure the incremental value relevance.

Subsequently, Odoemelam et al. (2019) use the dummy variable to create interaction coefficients as independent variables with the purpose of investigating the dummy’s interaction with BVPS. Following this adjustment, in Model (2) a dummy variable is added the for pre and post the implementation of IFRS 16 (IFRS16), coding the pre period as 0 and the post period as 1. Moreover, interaction variables are added as independent control variables to analyze the interaction between IFRS 16 and lnSIZE (lnSIZE*IFRS16), the interaction between EBITDA and IFRS 16 (EBITDA*IFRS16) and the interaction between LEV and IFRS 16 (LEV*IFRS16). Adding the interaction variables forces the variables lnSIZE, LEV and EBITDA to be dependent on the IFRS 16 implementation and thus allows to measure the incremental value relevance effect. The interaction variables are indicating if IFRS 16 influences the relationships of lnSIZE, LEV and EBITDA with MVPS which is contributing to validate which variables potentially contribute to affect value relevance.

Model (2) Incremental value relevance model

MVPSit = αi + β1BVPSit + β2EBITDAit + β3lnSIZEit + β4LEVit + β5IFRS16 + β6EBITDAit*IFRS16 + β7lnSIZEit*IFRS16 + β8LEVit*IFRS16 + β9DQ1 + β10DQ2 + β11DQ3 + εi

Market value per share MVPSit Measured by the market value of the firm’s share price, at end of three months following the book end date of the financial quarter- end, at time t and firm i

+

Earnings before interest, taxes,

depreciation and amortization EBITDAit Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares, at the book end date for the financial quarter-end, at time t and firm i

+

Size lnSIZEit Natural log of total assets, at the book end

+

-

IFRS 16 dummy IFRS16 Dummy variable where the pre IFRS 16

period is coded = 0, and the post IFRS 16 period is coded = 1

+

Interaction coefficient between

EBITDA and IFRS 16 EBITDAit*IFRS16 The interaction between the IFRS 16 dummy

variable, and EBITDA +

lnSIZE and IFRS 16 ^lnSIZEit*IFRS16 The interaction between the IFRS 16 dummy

variable and lnSIZE +

LEV and IFRS 16 LEVit*IFRS16 The interaction between the IFRS 16 dummy

variable and LEV -

effects +

Model (3) is based on the same foundations as Model (1) but is designed to test hypothesis II. Hence, Model (3) has additional industry-adjustments. For Model (3), the full sample is categorized into two groups:

(16)

16

“Consumer discretionary” representing 48 observations and “Non-Consumer discretionary” representing 214 observations. By adding an industry dummy variable (COND), Model (3) allows examination of value relevance on an industry level. We code “Non-Consumer discretionary” as 0 and “Consumer discretionary” as 1. We choose the industry group “Consumer discretionary” defined by Bloomberg Industry Classification Standard (2019), level 1, to be specifically studied. “Consumer discretionary” includes the BICS level 2 subsectors:

“Retail discretionary”, “Gaming, lodging and restaurants” which includes hotel corporates, and “Passenger transportation” which includes airline corporates (see appendix). Hence, the BICS industry group classification

“Consumer discretionary” includes all the relevant industries according to previous research stating that the sectors most affected by lease capitalization are retail, transportation, airlines, and hotels (Gosman & Hanson 2000; Morales-Díaz & Zamora-Ramírez, 2018). Similarly, to Model (2), we add interaction variables to examine if lease intensity influences the relationship between our primary variables and market value. Hence, our interaction variables are lnSIZE interacting with lease intense industries (lnSIZE*COND), the interaction between EBITDA and lease intense industries (EBITDA*COND) and the interaction between LEV and lease intense industries (LEV*COND).

Model (3) Relative value relevance model, with incremental analysis of industry

Pre IFRS 16-implementation period

MVPSit = αi + β1BVPSit + β2EBITDAit + β3lnSIZEit + β4LEVi + β5COND + β6EBITDA*COND + β7lnSIZE*COND + β8LEV*COND + β9DQ1 + β10DQ2 + β11DQ3 + εi Post IFRS 16-implementation period

MVPSit = αi + β1BVPSit + β2EBITDAit + β3lnSIZEit + β4LEVit + β5COND + β6EBITDA*COND + β7lnSIZE*COND + β8LEV*COND + β9DQ1 + β10DQ2 + β11DQ3 + εi

Market value per share MVPSit Measured by the market value of the firm’s share price, at end of three months following the book end date of the financial quarter- end, at time t and firm i

+

Earnings before interest, taxes, depreciation and amortization, per share

EBITDAit Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares, at the book end date for the financial quarter-end, at time t and firm i

+

Size

^lnSIZEit Natural log of total assets, at the book end

+

-

Consumer discretionary dummy IFRS16 Dummy variable where the Non-Consumer discretionary sectors is coded 0, and the Consumer discretionary sectors is coded 1

+

Interaction coefficient between EBITDA and COND

EBITDAit*COND The interaction between the COND dummy variable, and EBITDA

+

lnSIZE and COND ^lnSIZEit*COND The interaction between the COND dummy

variable and lnSIZE +

LEV and COND LEVit*COND The interaction between the COND dummy

variable and LEV -

effects. +

(17)

17

Coherently with Ohlson (1995), we are observing the adjusted R² metric closely over time, following the pre-implementation period and the post-implementation period. The adjusted R² metric measures the relative value relevance effect, thereby, if adjusted R² has increased in the post-period, hypothesis II is true.

4.3 Specification tests

A univariate analysis is conducted by studying the mean values of the variables with the purpose of identifying differences in the characteristic of the sample between the two periods. We conduct a paired sample t-test to test for significant differences between the pre and post period. Furthermore, we conduct a bivariate analysis using the Pearson correlation matrix for the pre and post period respectively, to observe any differences in correlation. We test for multicollinearity using the Variance Inflation Factor (VIF) and tolerance values.

We use panel data for all our regressions and test which model is the most consistent with our sample.

We run a pooling model regression and within model regressions, with fixed effects as time and individual effects, two-way effects and random effects. F-tests are used to test for any significant effects, between the pooling attribute and fixed effects of time, individuality and two-way, to find the most consistent model for our sample. The Hausman test is conducted to test whether to use the fixed effect model or the random effect model.

4.4 Diagnostic tests for autocorrelation, heteroscedasticity and cross-sectional dependence

Common issues with panel data are the presence of autocorrelation, heteroscedasticity and cross-sectional dependence (Mirza et al., 2019). First, we perform a Wooldridge test to detect autocorrelation in the sample.

Secondly, we conduct a Breusch-Pegan tests and a King Wu tests for all models, to test whether heteroscedasticity is present. Lastly, we perform a Pesaran test to find possible cross-sectional dependence in our models. To adjust the results when using data diagnosed with mentioned characteristics, Beck and Katz (1995) recommend estimating linear models of panel data by OLS, using a sandwich type estimator of the covariance matrix of estimated parameters called Panel Correction Standard Error (PCSE). The PCSE estimator provides robust standard errors, also in the cases when N, the panels cross sectional dimension, is greater than T, the time dimension, which is an issue for other methods such as FGLS (Parks, 1967). The PCSE sandwich estimator has been employed by several recent value relevance studies within the field of accounting (Mirza et al., 2019). As our models are based on the basic formation of the model by Mirza et al. (2019), using market value per share, book value per share, EBITDA per share, the ratio of total debt to asset and the natural logarithm of total assets, our models could contain heteroscedasticity, autocorrelation and cross-sectional dependence.

Thus, in line with Mirza et al. (2019), the PCSE approach recommended by Beck and Katz (1995) is adopted to adjust for the mentioned issues.

(18)

18

5. Empirical results

5.1 Univariate analysis

Table I. Paired Sampled T-test

Pairs Mean difference Std. Deviation Std. Error Mean t Sig. (2-tailed)

Pair 1 MVPS - MVPS' -8.282 44.858 1.385 -5.977 0.000

Pair 2 BVPS - BVPS' -3.765 12.459 0.384 -9.783 0.000

Pair 3 EBITDA – EBITDA' -181.828 1167.116 36.052 -5.043 0.000

Pair 4 lnSIZE - lnSIZE’ -0.133 0.289 0.008 -14.928 0.000

Pair 5 LEV - LEV' -0.042 0.116 0.0036 -11.886 0.000

Note: MVPS: Market value per share three months after the book end date before the IFRS 16 adoption; MVPS': Market value per share three months after the book end date after the IFRS 16 adoption; BVPS: Book value of equity per share at the book end date before the IFRS 16 adoption; BVPS':

Book value of equity per share at the book end date after the IFRS 16 adoption; EBITDA: Ratio of Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares at book end date before the IFRS 16 adoption; EBITDA': Ratio of Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares at book end date after the IFRS 16 adoption; lnSIZE: Natural log of total assets at book end date before the IFRS 16 adoption; lnSIZE': Natural log of total assets at book end date after the IFRS 16 adoption; LEV: Ratio of total debt to total assets at book end date before the IFRS 16 adoption; LEV': Ratio of total debt to total assets at book end date before the IFRS 16 adoption

The purpose of the univariate analysis is to test if the characteristics of our sample indicate significant changes in the post period of the implementation of IFRS 16 compared with the pre period. Table I compare the means of the variables through a paired sample t-test. The results indicate that the variables are significantly different in the post period of the adoption. All variables have a statistically significant higher mean value post IFRS 16 than before as p < 0.001. In line with the prediction by IASB (2016), EBITDA, total asset, and leverage has increased, as the mean value for the post period is significantly higher compared to the pre period when conducting a paired sample t-test. Table I summarizes the paired sample descriptive statistics.

5.2 Bivariate analysis

Table II. Pearson correlation matrix

Pre IFRS 16 MVPS BVPS EBITDA lnSIZE LEV

MVPS 1

BVPS 0.490** 1

EBITDA 0.259** 0.219** 1

lnSIZE 0.370** 0.511** 0.495** 1

LEV 0.095** 0.149** 0.098** 0.416** 1

Post IFRS 16

MVPS’ 1

BVPS’ 0.507** 1

EBITDA’ 0.251** 0.243** 1

lnSIZE’ 0.374** 0.514** 0.478** 1

LEV’ 0.034 0.048 0.030 0.358** 1

** Correlation is significant at the 0.01 level (2-tailed).* Correlation is significant at the 0.05 level (2-tailed).

Note: MVPS: Market value per share three months after the book end date before the IFRS 16 adoption; MVPS': Market value per share three months after the book end date after the IFRS 16 adoption; BVPS: Book value of equity per share at the book end date before the IFRS 16 adoption; BVPS':

Book value of equity per share at the book end date after the IFRS 16 adoption; EBITDA: Ratio of Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares at book end date before the IFRS 16 adoption; EBITDA': Ratio of Earnings before interest, taxes, depreciation and amortization, divided by fully diluted outstanding shares at book end date after the IFRS 16 adoption; lnSIZE: Natural log of total assets at book end date before the IFRS 16 adoption; lnSIZE': Natural log of total assets at book end date after the IFRS 16 adoption; LEV: Ratio of total debt to total assets at book end date before the IFRS 16 adoption; LEV': Ratio of total debt to total assets at book end date before the IFRS 16 adoption

(19)

19

Table II summarizes the correlations between all our variables. The Pearson correlation matrix, both pre and post IFRS 16, shows that our dependent variable MVPS is significantly positively correlated on the 0.01 level with BVPS, EBITDA, and lnSIZE. The only variable not being significant on the 0.01 significance level in both periods is LEV, showing a lower correlation in the post period together with a change from 0.01 significance level to being insignificant in the post period. The correlation matrix shows that all correlations between MVPS and the independent variables have a similar relationship in the post IFRS 16 period as in the pre period.

We are checking for collinearity to assure multicollinearity is not an issue in our sample. The VIF values for our independent variables with the results of multicollinearity is non-existent as the VIF values are below the threshold of 10 and the tolerance values are above 0.1 (Pallant, 2016).

5.3 Multivariate analysis

We adopt econometric tools to measure the validity of all four models. First, we confirm that all models, including both time periods, are globally significant (P-value = 0.000). Secondly, the models are tested to determine whether model attribute of pooling, random or within fixed effects, in the formats of time, individual or two-way, is the most consistent model attribute for our sample. The F-tests compares the pooling attribute with the fixed effects, showing that fixed effects are significantly present for all models. The F-test comparing the two-way attribute versus the time effects attributes shows that time is present for all models. However, when the time attribute and two-way attribute is applied, the time dummies of seasonal effects (DQ1, DQ2, DQ3) included in our models and the IFRS 16 dummy (IFRS16) included in Model 2, are dropped due to collinearity.

This is due to the time effect is handled by the dummies and therefore both the format of time and two-way effects are not applicable. Therefore, the fixed effect with individual format is the most consistent attribute for our sample as that allows to deal with the time effect through the dummies only and not through the model attribute. Hence, through the fixed effect with individual format, there is not an issue of collinearity and the time dummies are not dropped in the models. We run a Hausman test to find whether fixed effect or random effects are our most consistent model attribute, whereas the p-value of 0.000 suggests the fixed effect attribute.

Therefore, the specification tests conclude that the most consistent model for our sample are fixed effects in the format of individual effects.

Furthermore, the Wooldridge which test for autocorrelation show that existence of autocorrelation is present in our sample as p-value < 0.05 suggests serial correlation. The Breusch-Pagan test and the King Wu test which checks homogeneity shows that the null of no homoscedasticity is rejected, suggesting that heteroscedasticity is present for all models with the alternative hypothesis of significant effects present. Lastly, the Pesaran test show existence of cross-sectional dependence as p-value < 0.001. As our sample includes autocorrelation, heteroscedasticity and cross-sectional dependence our results are adjusted using PCSE as recommended by Beck and Katz (1995).