Implications of IFRS 16 adoption Evidence from Swedish publicly listed firms

(1)

Implications of IFRS 16 adoption

Evidence from Swedish publicly listed firms

Master’s Thesis 15 credits

Department of Business Studies Uppsala University

Spring Semester of 2020

Date of Submission: 2020-06-27

Jonathan Spånberger

(2)

Abstract

In this study, we investigate how the implementation of IFRS 16 is affecting the financial statements of Swedish publicly listed firms, and what implications there are for financial statement users. These effects are analyzed by looking at transitional effects on total assets, total liabilities and EBITDA and by comparing different sectors, following estimations of sectoral differences in prior studies (e.g. Fülbier et al., 2008; Morales-Díaz & Zamora-Ramírez, 2018a). As a way of approximating the practical implications of IFRS 16, this study is analyzing changes in the key financial ratios: D/E and EV/EBITDA.

We find significant median increases in total assets, total liabilities and EBITDA in the full sample, as well as within each sector group. Further, we confirm the existence of sectoral differences, finding the largest median increases in the Consumer Services sector and the smallest in the Financials sector. We also confirm that IFRS 16 bring new implications for financial statement users, since important and commonly used financial ratios are significantly changed: we observe a significant median increase in the D/E ratio and a significant median decrease in the EV/EBITDA multiple.

Keywords: IFRS 16, lease accounting, off-balance sheet financing, impact assessment,

(3)

1. Introduction

The aim of this study is to investigate how the implementation of IFRS 16 – the new accounting standard on leasing – is affecting the financial statements of Swedish publicly listed firms, and what implications there are for financial statement users. IFRS 16 is the most recently implemented standard within the International Financial Reporting Standards (IFRS): a series of global, principle-based, accounting standards aimed at facilitating cross-border capital movement by eliminating national differences in accounting regulations (IFRS Foundation, 2019a). The standards are used widely all over the world, including Sweden where IFRS has been mandatory for publicly listed firms since 2005 (European Commission, 2019).

IFRS 16 was implemented in fiscal years started January 1^st, 2019 or later. The main impact is that future payments of operating leases need to be capitalized. IFRS 16 replaced IAS 17, under which finance leases were capitalized but fees from operating leases were reported as operating expenses, complemented with information disclosed in notes. According to the IFRS Foundation, the aim of IFRS 16 is to “faithfully represent lease transactions” and to “provide a basis for users of financial statements to assess the amount, timing and uncertainty of cash flows arising from leases” (IFRS Foundation, 2019b). Hence, an implicit aim is to decrease the information asymmetry between the firms and their stakeholders (Hoogervorst, 2016). During the last decades, there seems to have been an increase in the usage of operating leases, creating a suspicion of companies deliberately using operating leases as a way of keeping assets and liabilities out of the financial statements, so called “off-balance sheet financing” (Abdel-Khalik, 1981; Imhoff & Thomas, 1988; Imhoff et al., 1991; Reason, 2005; Duke et al., 2009). Previous studies such as Imhoff et al. (1993) are pointing out how disclosed, but not recognized, information about operating leases in the annual reports (as was the case under IAS 17) seem to not be utilized to the same extent by all stakeholders. Hence, the results point to a need of including such information in the financial statements, i.e. capitalizing the operating lease expenses. In similar studies, although not explicitly about leasing, Ahmed et al. (2006) and Davis-Friday et al. (1999) also provide evidence of differences in utilization of the information in recognized and disclosed amounts.

There has, however, been a lot of criticism of the IFRS 16 and its counterparty within US GAAP: FASB ASC 842 (Accounting Today, 2013; Tysiac, 2013; Bratten et al., 2013; Altamuro et al., 2014). The criticism is not mainly regarding whether or not more information is needed,

(5)

proportion between benefits and negative economic consequences that could be caused by the new standard (Accounting Today, 2013; Tysiac, 2013; Kabureck, 2015). Some business representatives claim that capitalizing operating leases is a complex and time-demanding activity, which does not have proportionate benefits for stakeholders (Accounting Today, 2013). Further, Bratten et al. (2013), Altamuro et al. (2014) and Giner & Pardo (2018) point to evidence indicating that a lot of stakeholders are in fact utilizing disclosed and recognized information about operating leases in a similar way. Hence, a possibility of achieving similar results with a less complex accounting standard than IFRS 16 is indicated.

Concluding the debate on lease capitalization, there exist evidence pointing in different directions regarding the most appropriate way to decrease the information asymmetries that seem to exist under IAS 17 and US GAAP (e.g. Imhoff et al., 1993; Bratten et al., 2013;

Altamuro et al., 2014). However, rather than the studies showing explicitly contradicting evidence, the opposite arguments of the debate stem from conflicting assessments of what level of benefit is proportionate to the costs of increased accounting input. For instance, regarding if it is necessary for all, or just most, stakeholders to be able to utilize the information, and to what extent external parties can be required to make their own valuations of the information.

Previous studies are commonly using slightly varying estimation models based on the constructive capitalization model (Imhoff et al., 1991) to estimate the effects from capitalization of operating leases. Prior studies, investigating markets in North America (Mulford & Gram, 2007; Durocher, 2008; Duke et al., 2009), Europe (Fülbier et al., 2008; Branswijck et al., 2011;

Morales-Díaz & Zamora-Ramírez, 2018a, 2018b) and Australia (Wong & Joshi, 2015), all show significant effects on the financial statements, primarily increases in total asset and total liabilities. These effects have implications also on key financial ratios such as ROA and D/E and are in many studies not uniformly distributed across industries and sectors. For instance, Fülbier et al. (2008) identify retail and fashion as lease intensive sectors being particularly affected by lease capitalization. Similarly, Branswijck et al. (2011) identify manufacturing, and Morales-Díaz & Zamora-Ramírez (2018a) identify retail, hotels and transportation. Our study focuses on transitional effects in three measures in the financial statements: total assets, total liabilities and EBITDA. This is due to the main changes of the IFRS 16 adoption being an addition of right-of-use assets and lease liabilities, together with the replacement of operating lease expenses with depreciation and interest expenses. As a way of approximating the implications to practitioners, the study is also examining two additional measures: the leverage

(6)

Firms are often experiencing the process of implementing IFRS standards as complex and quite burdensome (Jermakowicz & Gornik-Tomaszewski, 2006). The presence of the preceding discussions and criticism (e.g. Accounting Today, 2013; Kabureck, 2015) suggests that many firms expect this to be true for IFRS 16 as well. Further, prior studies show that regulation aimed at harmonizing accounting standards internationally is not guaranteed to achieve its goal, just because the standards are uniform (Soderstrom & Sun, 2007; Holthausen, 2009).

Differences in outcome can still occur due to country-specific institutional and economic factors (ibid.). In this wider context, comparisons of country-specific settings and differences in implications of accounting standards are of great relevance in understanding the challenges of international harmonizing of financial reporting. In such comparisons, Sweden is an interesting country to analyze, due to its ability to be an appropriate representative for more than one group of countries. For instance, by having a relatively small economy, by being part of the European Union, and by being one of the top 10 most competitive countries in the world (IMD World Competitiveness Center, 2019; Schwab, 2019), it can be used in multiple relevant comparisons.

This study is relevant to users of financial statements due to its analysis of IFRS 16-caused changes in key financial ratios, and of sectoral differences. The findings can help these users to assess the appropriateness of different multiples and ratios in comparison across sectors and in retroactive analysis of firms. It also contributes to the academic fields of lease accounting and harmonization of financial reporting. Due to the complexity of IFRS 16 (Accounting Today, 2013; Tysiac, 2013; Altamuro et al., 2014) and the general difficulties of implementing new accounting standards (Jermakowicz & Gornik-Tomaszewski, 2006), our study contributes by examining the actual, reported numbers calculated by the companies themselves during the transition to IFRS 16. Due to the need of understanding implementation of new accounting standards in different countries and settings, our study also contributes both by analyzing the case of Sweden, and by making sectoral comparisons. Finally, this study adds to the debate on the usefulness and informational content of lease capitalization (e.g. Duke et al., 2009; Bratten et al., 2013) by investigating key financial ratios that are important to firms’ external stakeholders. In the study, two research questions are answered:

1. What effects did the implementation of IFRS 16 have on the financial statements of Swedish publicly listed firms?

2. Did the effects cause implications for financial statement users?

(7)

2. About IFRS 16

The IFRS are issued by the International Accounting Standards Board (IASB), overseen by the IFRS Foundation (IFRS Foundation, 2019c). The standards are used all around the world and firms listed on stock exchanges are in most countries required or permitted to use the IFRS standards. One important example is the EU where, since 2005, all publicly listed firms are required to use IFRS for the group consolidated numbers (European Commission, 2019). One important exception, still using national GAAP, is the USA. However, the US GAAP and the IFRS are ensured to be conformed through the “Norwalk Agreement” (FASB, 2002), hence IFRS should be seen as globally influential, even if not implemented in all major capital markets.

New IFRS standards are issued on a regular basis to replace previously used International Accounting Standards (IAS). IFRS 16 is the most recently implemented standard, issued to replace IAS 17, with an objective of ensuring high quality and transparency of lease accounting in financial statements (Hoogervorst, 2016). IFRS 16 was issued in January 2016 and the standard applies to fiscal years starting January 1^st, 2019 or later.

IFRS 16 is an accounting standard on leasing. Two forms of leasing exist: operating and finance leasing, which were treated differently under IAS 17. An operating lease is a contract that allows a lessee use of an asset without transferring ownership or risks related to the asset. A finance lease is a contract that allows a lessee use of an asset without transferring ownership, but where the risks and rewards are transferred so that the transaction is very close to a purchase transaction. To achieve the aim of IFRS 16 – to “faithfully represents lease transactions” and

“provide a basis for users of financial statements to assess the amount, timing and uncertainty of cash flows arising from leases” (IFRS Foundation, 2019b) – lessees are now required to treat operating and finance leasing similarly. This is performed through the recognition of a lease liability and a right-of-use asset in the balance sheet. Thereby, IFRS 16 eliminates the difference between operating leases and finance leases, requiring lessees to recognize assets and liabilities for all lease contracts. However, the standard is voluntarily applied for those contracts where the lease term is less than 12 months or the contract has a low value (IFRS Foundation, 2019b).

Under IAS 17, information about operating lease expenses, including future expenses from non- cancellable contracts, was required to be disclosed in notes in the annual reports. Under IFRS 16, firms are instead required to recognize a right-of-use asset and a lease liability in the balance

(8)

amount of the lease liability and other payments that include the lessee’s initial direct cost, prepayments and estimated restoration obligations (IASB, 2016). The lease liability is determined at the present value of the lease payments payable over the lease term. The present value is calculated as the non-cancellable lease contracts, discounted with a discount rate. The discount rate used should be the “…interest rate implicit in the lease, if that rate can be readily determined. Otherwise the lessee shall use its incremental borrowing rate…” (IASB, 2016, p.

26). Due to this regulation, different discount rates can be used in different companies, although sectoral and industrial similarities could be expected, following that companies within an industry could be facing similar risks and financing situations.

Following the changes in the balance sheet, the profit and loss statement is adjusted by replacing the operating lease expenses with depreciation of the right-of-use asset and interest expenses based on the size of the lease liabilities. This implies that P&L changes exist primarily in EBITDA. Cash flow statements are adjusted as well, since the lease payments are now not comprising operating expenses but are classified as repayments of the liability and interest (i.e.

financial) expenses (IFRS Foundation, 2019b). In contrast to lessee accounting, lessor accounting remains the same under IFRS 16. Lessors are able to classify a lease as an operating lease or a finance lease. When a lease allows transferring all risks and rewards of the ownership of an underlying asset, it is classified as a finance lease. Otherwise, it will be classified as an operating lease.

(9)

3. Theory and literature review

3.1 Asymmetric information and IFRS

Information plays an important role in all financial decision-making. In neo-classical economics, a situation of perfect information is often assumed. In contrast to this theoretical assumption, a common situation in different kinds of markets seems to be that all participants do not, in fact, have access to equal information, i.e. a situation of asymmetric information (Akerlof, 1970; Spence, 1973; Stiglitz, 1975). In financial markets, there is a general fear of asymmetric information since it can cause problems such as adverse selection or moral hazard (Akerlof, 1970). The effects on trade and transactions caused by information asymmetries are assessed to be harmful for capital markets and society in general and are therefore counteracted with regulation such as accounting standards (Assidi & Omri, 2012; Kao & Wei, 2014).

Accounting standards in particular are aimed at addressing the asymmetric information in the relationship of company management and shareholders, often referred to as the principal-agent problem (Jensen & Meckling, 1976), where neither information nor interests are automatically aligned between parties. Accounting standards, however, are just one aspect of dealing with asymmetric information, since it is also affected by a firm’s corporate governance model, the work performed by auditors, country-specific legal aspects etc. (Jensen & Meckling, 1976).

Previous studies suggest that adoption of a uniform set of accounting standards, such as the IFRS, can mitigate information asymmetry and improve accounting quality (Barth et al., 2008;

Daske et al., 2008; Muller et al., 2011). Positive effects such as increased market liquidity and decreased cost of capital (Daske et al., 2008) as well as decreased earnings management and increased value relevance of accounting (Barth et al., 2008) have been shown using large international samples. In more specific areas, positive effects have also been shown in aspects such as reduced underpricing of IPO: s¹ (Leone et al., 2007; Hong et al., 2014) and by decreased information asymmetry in the case of the European real estate industry (Muller et al., 2011).

New accounting standards, such as IFRS, are in general aimed at increasing accounting quality and are expected to minimize barriers to cross-border trading in securities. This can in turn increase market efficiency and reduce the cost of capital (Daske et al., 2008). Improved transparency, comparability and quality of financial reporting have positive effects such as helping investors to make more efficient investment decisions. However, as previous studies

(10)

show (Jermakowicz & Gornik-Tomaszewski, 2006; Soderstrom & Sun, 2007; Holthausen, 2009), it is not always the case that these kinds of expectations are met when implementing new accounting standards. Jermakowicz & Gornik-Tomaszewski (2006) examine the difficulties in implementing IFRS from the perspective of EU publicly traded companies. The results indicate that most companies identify some limitations regarding the process of implementing IFRS, e.g. it being a costly, complex and burdensome process. Also, these preparers of financial statements claim to face problems in the implementation process due to a lack of implementation guidance and uniform interpretations (Jermakowicz & Gornik-Tomaszewski, 2006).

The argument of increased market efficiency is based on the assumption that a uniform set of accounting and reporting standards will work in the same way wherever it might be implemented. However, some prior studies (Soderstrom & Sun, 2007; Holthausen, 2009) question whether it is sufficient to implement uniform standards to achieve comparability of financial statements between different countries. For instance, Soderstrom & Sun (2007) show that accounting quality differ across countries, which is explained due to institutional differences, and thus conclude that differences in accounting quality are likely to remain after IFRS adoption. Holthausen (2009) similarly suggests that country-specific institutional settings might have an even more important impact on the financial reporting outcome than the reporting standards themselves.

3.2 Asymmetric information and lease accounting

Companies across different industries have different strategies of lease usage. Certain industries might experience larger benefits from lease usage than others, which causes sectoral differences in lease intensity (Smith & Wakeman, 1985; Finucane, 1988; Adams & Hardwick, 1998).

According to Finucane (1988), companies in air transport and retail industries use more lease financing than others. Adams & Hardwick (1998) find that companies in service and utilities sectors use more leases, while for example construction companies have a low tendency to lease.

For a long time, there has been a trend of increased use of operating leases, while the same development cannot be seen in finance leasing (Abdel-Khalik, 1981; Imhoff & Thomas, 1988;

Duke et al., 2009). Due to the different accounting treatment of finance and operating leasing, the trend of increased operating leasing has been commonly interpreted as companies

(11)

“off-balance sheet financing” (Abdel-Khalik, 1981; Reither, 1998; Duke et al., 2009). Off- balance sheet financing can have a favorable effect on different financial measures, e.g. debt- to-equity leverage (D/E), enterprise value (EV) and return on assets (ROA). Hence, a strategy of off-balance sheet financing has the potential to create large information asymmetries between company management and different stakeholders, why the arguments for similar accounting treatment of all leasing have grown stronger (Imhoff et al., 1993; Duke et al., 2009).

Previous studies investigate off-balance sheet financing by estimating the effects from lease capitalization and by analyzing how information disclosed in notes respectively recognized in the balance sheet is used by stakeholders (e.g. Imhoff et al., 1993; Duke et al., 2009; Giner &

Pardo, 2018). Imhoff et al. (1991, 1997) introduce a model for estimating the effects on the financial statements called the constructive capitalization model². Using this model, they examine how investors respectively executive compensation committees in the USA seem to utilize disclosed information differently (Imhoff et al., 1993), which points towards an existence of information asymmetries. Their findings suggest that investors are utilizing the disclosed information of operating leases when assessing the riskiness of a firm’s shares, but no evidence is found of executive compensation committees adjusting reported numbers with disclosed information about operating leases when establishing CEO compensation (Imhoff et al., 1993).

Duke et al. (2009) also argue that companies, by using operating leases, can hide large amounts of liabilities and assets from investors, as well as report more favorable net income, hence creating large information asymmetries.

Further, Ahmed et al. (2006) and Davis-Friday et al. (1999) also show evidence of that disclosed information is not necessarily utilized to the same extent as recognized information, however not specifically examining information on leasing. Analyzing information on fair value of derivative financial instruments in US companies’ reporting, Ahmed et al. (2006) present evidence of significant valuation coefficients on recognized derivatives, but not on disclosed derivatives. Their conclusion is therefore that recognized information contains more utilizable information than disclosed information, thereby suggesting that disclosure and recognition is not to be seen as substitutes. Same conclusions are drawn by Davis-Friday et al. (1999) using information on anticipated liabilities for retiree benefits in US companies.

2 A model for estimating the size of the effects on assets, liabilities and net income from capitalization of the noncancelable commitments embodied in a firm's operating leases. The model was developed by Imhoff et al.

(12)

However, there are also prior studies showing somewhat contradicting findings, indicating that information in disclosed amounts is not per se less utilized than in recognized amounts. Bratten et al. (2013) suggest that market participants in the US capital markets do in fact utilize disclosed and recognized information on operating leases similarly, as long as the information in disclosed items is reliable and not based on management estimates. Similar results are found by Giner & Pardo (2018) for Spanish firms in the retail sector. Further evidence on similar treatment of recognized and disclosed information is provided by Altamuro et al. (2014) in a study of US bank loans, showing that credit agencies are utilizing the disclosed information on operating leases. This implies that, for companies with credit ratings, banks deciding on whether to issue a loan or not can access the information on operating leases without it being recognized (Altamuro et al., 2014).

Concluding this debate on whether or not capitalization of operating lease expenses is needed and valuable, there is some contradicting evidence regarding how different stakeholders are utilizing disclosed information (e.g. Davis-Friday et al., 1999; Ahmed et al., 2006; Bratten et al., 2013; Altamuro et al., 2014). Even so, there are indications of an existence of some information asymmetries following evidence of not all stakeholders being able to utilize the disclosed information uniformly (Imhoff et al., 1993). However, the level of information asymmetry of course also depends on aspects such as the specific firm’s governance model, and legal and institutional factors (Jensen & Meckling, 1976).

3.3 Effects of capitalizing operating leases

Even though IFRS 16 was first implemented in 2019, there are a multitude of prior studies estimating the effects from capitalization of operating lease expenses (e.g. Imhoff et al., 1991, 1997; Durocher, 2008; Fülbier et al., 2008; Duke et al., 2009; Wong & Joshi, 2015; Morales- Díaz & Zamora-Ramírez, 2018a, 2018b). These studies are performed in different national settings and are based on information disclosed in notes in the financial statements, showing significant effects on financial statements and different financial ratios. The following sections summarize the findings of recent studies on North American, Australian and European markets, compiled by balance sheet effects and profit and loss statement effects.

3.3.1 Balance sheet effects

Due to the aim of lease capitalization being to reduce off-balance sheet financing, the primary anticipated effects are on balance sheet items. This is estimated in studies using adjusted

(13)

versions of the constructive capitalization model (Imhoff et al., 1991) on North American, Australian as well as European capital markets.

In a large study of US companies from the S&P index, Duke et al. (2009) show a significant mean increase in total liabilities (11.13%) and total assets (3.97%). The study also analyzes ratios such as D/E, showing an increase in the mean value of approximately 13% (Duke et al., 2009). Durocher (2008) shows similar results in a study on large Canadian companies, with median increases in total liabilities (6.0%) and in total assets (2.6%). Durocher (2008) further shows findings such as increased D/A ratio and decreased current ratio, following the balance sheet effects (Durocher, 2008), and show indications of some sectoral differences. In an industry specific study, using a lease intensive sample of US retail companies, Mulford & Gram (2007) show significant increases, larger than the studies with broader samples (e.g. Duke et al., 2009; Durocher, 2008), with median increases in total liabilities (26.4%) and total assets (14.6%). A median increase in the D/E ratio (26.4%) is also emphasized, as well as a decrease in ROA (-1.7%) (Mulford & Gram, 2007).

In a study of the Australian capital market, using a sample of 107 companies, Wong & Joshi (2015) identify significant effects in the form of a mean increase in total liabilities (4.34%), a mean increase in total assets (3.47%) and a mean decrease in total equity (-0.27%). Financial ratios such as D/E and ROA are analyzed as well, showing a relative increase in the mean value of D/E (31.69%), and a relative decrease in the mean value of ROA (-15.35%).

In the European context, Fülbier et al. (2008) investigate a German sample of 90 companies from three major German indices, showing a median increase in total liabilities of 17.3% and in non-current assets of 8.5%. Financial ratios such as D/E and ROA are further analyzed, with a median change in relative terms of 8.0% respectively -0.3%. The study further identifies retail and fashion industries as particularly affected by the lease capitalization, and for example natural resources and energy as quite unaffected (Fülbier et al., 2008). Compared to other studies from the same period of time, Fülbier et al. (2008) seem to present effects of a rather large magnitude, something that is explained by the usage of individual discount rates for each company, lower than the uniform rates used by many other studies (e.g. Duke et al., 2009;

Branswijck et al., 2011).

Branswijck et al. (2011) investigate country-specific differences in a study of lease capitalization in Belgium and the Netherlands. For the total sample, including companies from

(14)

both countries, mean increases in total liabilities of 5.80% and in total assets of 3.00% are identified, as well as an increase in the mean value of D/E ratio of 8.4%, a decrease in the mean value of current ratio of -3.5%, and no material changes in ROA (Branswijck et al., 2011). In order to test country-specific differences, Branswijck et al. (2011) perform a regression analysis with the estimated lease liability as the dependent variable and country and industry as two of the independent variables. They are able to show significant impact from both industry and country (Branswijck et al., 2011).

Morales-Díaz & Zamora-Ramírez (2018a, 2018b) are investigating lease capitalization effects on a sample of European respectively Spanish companies, using a model based on the final version of IFRS 16 instead of a more general constructive capitalization model. In a study of more than 600 European companies (2018a), Morales-Díaz & Zamora-Ramírez show effects of a median increase of 11.2% in total liabilities and of 5.2% in total assets. In a similar study of Spanish companies (2018b), median increases in total liabilities respectively total assets are shown of 6.5% respectively 3.8%

Morales-Díaz & Zamora-Ramírez (2018a) also show a median increase in the D/E ratio of 14.9%, a median increase in the D/A ratio of 4.9% and actually a median increase in ROA of 0.7%. Morales-Díaz & Zamora-Ramírez (2018b) show both similar and different effects for the Spanish sample: a median increase in D/A of 2.5% but a median decrease in ROA of -1.7%

(however the mean value is an increase). Both studies (2018a, 2018b) identify significant differences between sectors. Balance sheet effects are particularly intense in Retail and Hotels (2018a, 2018b) as well as Foods and Transportation (2018a), while Financial and Real Estate companies experience the smallest changes (2018a, 2018b).

3.3.2 Profit and loss statement effects

Due to the balance sheet effects of increased assets and liabilities, effects on EBITDA follow from the replacement of operating lease expenses with depreciation and interest expenses.

Recent studies (e.g. Mulford & Gram, 2007; Fülbier et al., 2008; Duke et al., 2009; Morales- Díaz & Zamora-Ramírez, 2018a, 2018b) investigate effects on the profit and loss statements following the example of Imhoff et al. (1997), but what measures being analyzed (EBITDA, EBIT, Net income, or included in ratios such as ROA) vary between studies. In this section, findings from studies presenting profit and loss statement effects, together with literature pointing to practical implications, are presented.

(15)

In the US setting, Duke et al. (2009) report P&L effects divided by positive and negative income subgroups: a mean increase of Net income for the negative income subgroup of 3.59% and a mean decrease for the positive income subgroup of -5.12%. Mulford & Gram (2007), in their study of US retail companies, do not report effects on Net income, but are instead reporting a median increase of EBITDA of 22.5%. In the European setting, Fülbier et al. (2008) investigate the P&L effects on EBIT in German companies, reporting a median increase of 2.9%, and investigate effects on the earnings-related valuation multiple P/E, without observing any significant effects. The studies on European and Spanish samples (Morales-Díaz & Zamora- Ramírez, 2018a, 2018b) are using both EBIT and EBITDA in the calculations of the ratios ROA and Financial expenses coverage, mentioning increases of EBITDA but not reporting the magnitude of the profit and loss statement effects separately.

Following the estimated effects in the profit and loss statement, as well as the effects in the balance sheet mentioned earlier, financial ratios and multiples using these measures as input would be also affected by the IFRS 16 adoption. One such multiple is the valuation multiple EV/EBITDA, affected by changes in EBITDA and total liabilities. According to Pinto et al.

(2019), performing a large global scientific survey on professional equity analysts to cover equity valuation practices, EV/EBITDA is together with P/E the most commonly used valuation multiple among practitioners. Loughran & Wellman (2011) further stress the practical importance of the EV/EBITDA multiple by showing empirical evidence of the multiple being able to explain realized stock returns, thus arguing that the common use of the multiple by practitioners is justified.

3.4 Hypotheses development

Based on the theory of information asymmetry and agency problems (Akerlof, 1970; Spence, 1973; Stiglitz, 1975; Jensen & Meckling, 1976), new accounting standards such as the IFRS are issued to decrease information asymmetries (Hoogervorst, 2016; IFRS Foundation, 2019a, 2019b). Following the attention that lease accounting and off-balance sheet financing have drawn in the last decades, IFRS 16 was, even though heavily criticized, implemented in 2019, requiring operating leases to be recognized in the balance sheet. Recent literature on North American, European and Australian markets (e.g. Duke et al., 2009; Fülbier, 2008; Wong &

Joshi, 2015) estimate increases in total liabilities, total assets and EBITDA following the capitalization of operating leases. Some studies also indicate that the effects from the final IFRS 16 standard might be even larger than the effects estimated using older capitalization models

(16)

(Morales-Díaz & Zamora-Ramírez, 2018a, 2018b). In order to analyze the effects on the financial statements of Swedish publicly listed companies, we formulate our first hypothesis:

H1: The implementation of IFRS 16 increased total liabilities, total assets and EBITDA of Swedish publicly listed firms.

Prior research shows differences in the usage of leasing in different sectors (Smith & Wakeman, 1985; Finucane, 1988; Adams & Hardwick, 1998). Recent literature also shows that different sectors exhibit largely different lease intensity when it comes to operating leases, and estimate larger effects from capitalization in some sectors than others; retail and hotels being the most commonly identified lease intensive industries (Mulford & Gram, 2007; Fülbier, 2008;

Morales-Díaz & Zamora-Ramírez, 2018a, 2018b). In this context, we formulate a second hypothesis:

H2: The implementation of IFRS 16 affected total liabilities, total assets and EBITDA of Swedish publicly listed firms differently across sectors.

It has been debated among practitioners (e.g. Accounting Today, 2013) as well as in academia (e.g. Duke et al., 2009) whether or not lease capitalization is contributing to external users of financial statements by enabling utilization of previously hidden information. While there is evidence of some financial statement users in fact being able to utilize information disclosed in notes similarly to information being recognized (Bratten et al., 2013; Altamuro et al., 2014;

Giner & Pardo, 2018), there is also evidence indicating that this is not the case (Ahmed et al., 2006; Davis-Friday et al., 1999), as well as research showing that not all stakeholders are able to utilize disclosed information about leasing uniformly (Imhoff et al., 1993). Following that the effects investigated in hypotheses 1-2 would cause changes to the input in key financial ratios, known to be commonly used by practitioners in risk assessment (Imhoff et al., 1993;

Fülbier et al., 2008) and company valuation (Loughran & Wellman, 2011; Pinto et al., 2019), we formulate a third hypothesis:

H3: The implementation of IFRS 16 caused changes in the D/E ratio and EV/EBITDA multiple of Swedish publicly listed firms.

(17)

4. Methodology

This study investigates the IFRS 16 effects on financial statements and key financial ratios of Swedish publicly listed firms. IFRS 16 was implemented in fiscal years starting in 2019, meaning that companies that do not use calendar year as fiscal year are in some cases not presenting their first full year of implementing IFRS 16 until late 2020 (i.e. after this paper is presented). However, in the last annual reports using IAS 17, all companies are required to disclose the transitional numbers of going into IFRS 16 reporting: the effects caused by addition of a right-of-use asset and a lease liability in the opening balances of the next fiscal year. In order to achieve the aim of the paper, we analyze the effects from IFRS 16 by examining the moment of transition from IAS 17 to IFRS 16.

4.1 Research design

Based on the underlying aim of IFRS 16 to reduce off-balance sheet financing, and through analysis of recent literature on lease capitalization (e.g. Duke et al., 2009; Morales-Díaz &

Zamora-Ramírez, 2018a), we have identified three primary financial measures that capture the IFRS 16 effects on the financial statements: total liabilities, total assets and EBITDA. Total assets and liabilities are directly affected by the inclusion of a right-to-use asset and a lease liability, and EBITDA is affected from the replacement of the prior operating lease expenses, included in EBITDA, with the depreciation of right-of-use assets and the interest expenses from the lease liability, both excluded from EBITDA. Analyzing the effects on these three measures in the full sample respectively by sector enable us to answer the first and third hypothesis of the study. Thus, these measures imply a valid method of achieving the aim of the study.

In order to approximate the implications to financial statement users, we will also calculate effects on a leverage ratio (D/E) and a valuation multiple (EV/EBITDA), two ratios directly affected by the changes in the financial statements. Due to its relevance in credit ratings and risk assessments, measuring companies’ ability to cover outstanding debt with shareholder equity, the effect on the D/E ratio is analyzed in prior research on lease capitalization (Durocher, 2008; Fülbier et al., 2008; Duke et al., 2009; Wong & Joshi, 2015; Morales-Díaz & Zamora- Ramírez, 2018a). The EV/EBITDA multiple is, according to Pinto et al. (2019), together with P/E the most commonly used multiple valuation models. Since the IFRS 16 adoption is expected to increase both EV (following increased liabilities) and EBITDA, this valuation multiple would also be affected (PwC, 2019). By using one measure commonly used in prior studies together with one additional measure identified to be practically relevant and likely to

(18)

be affected, we assess our method of approximating the implications to financial statement users to be valid and contributing to the field of lease accounting research.

4.2 Data and sample

The sample of this study is taken out of the total population of firms publicly listed on the Nasdaq OMX Stockholm main market – including Large Cap, Mid Cap and Small Cap – as of 2019-12-31. The population as of 2019-12-31 is identified to be the most relevant population due to that most companies are reporting full year effects of IFRS 16 in fiscal years ending at that date.

Out of a total population of 338 companies, we identify 59 companies needed to be excluded in order to get a sample where all companies share important institutional settings that respond to the aim of this study. These 59 companies are excluded out of six different aspects: being delisted or declared bankruptcy in early 2020 (thus losing future relevance to analysts), not being listed at the date of transition into IFRS 16, not being required to implement IFRS 16 due to not reporting consolidated numbers, having a non-Swedish parent companies, or being an investment company classified to the Financials sector (these companies can consolidate the numbers of group companies from a multiple of sectors, hence making analysis of sectoral differences blurry). Following these exclusions, the full sample consists of 279 companies and is distributed according to table 1.

Table 1: Sample exclusions

Large Cap Mid Cap Small Cap Total

Total population 100 138 100 338

Exclusion^1-4 2 7 5 14

Exclusion⁵ 14 13 3 30

Exclusion⁶ 7 6 2 15

Full sample 77 112 90 279

Notes: 1: Companies delisted in January 2020 (before publishing annual report 2019), 2: Companies declaring bankruptcy early 2020, 3:

Companies not listed 2019-01-01 (the date of the IFRS 16 implementation), 4: Companies not implementing IFRS 16 due to no consolidated reporting, 5: Groups having a non-Swedish parent company, 6: Investment companies categorized as Financials

This study is investigating effects from IFRS 16 and its implications to practitioners, as well as sector-specific effects. In order to analyze sectoral differences, the sample is categorized into 10 sectors using the Global Industry Classification Standard (GICS); the classifications are retrieved from Nasdaq. GICS is a globally used system for classification of companies into

(19)

Nordic as the Real Estate sector is grouped into the Financials sector (Nasdaq Group, 2020).

The usage of this global system in the study facilitates comparisons with studies in other countries. Since four of the sectors, however, only contained a few companies in our sample, we modified the GICS classifications by reclassifying the four smallest sectors into one larger sector group called Others (see table 2). After the reclassification, the smallest sector consists of 18 companies, which is assessed to be a sufficient sample size for the non-parametric tests performed in this study (Pallant, 2016, p. 183).

Table 2: Sector classification

Sector (GICS) Sector reclassification Large Cap Mid Cap Small Cap Total

Consumer Goods Consumer Goods 8 12 5 25

Consumer Services Consumer Services 7 14 8 29

Financials Financials 21 16 4 41

Health Care Health Care 6 23 22 51

Industrials Industrials 24 31 28 83

Technology Technology 2 15 15 32

Basic Materials

Others

6 0 5 11

Oil & Gas 1 1 1 3

Telecommunications 2 0 1 3

Utilities 0 0 1 1

Sum of all sectors 77 112 90 279

Note: The table consists of the number of firms in the sample divided by sector (before and after reclassification) and by market segment.

To retrieve the data needed to perform the calculations of the IFRS 16 effects, data is manually collected from annual and quarterly reports, the Nasdaq website (www.nasdaqomx nordic.com) and the database Retriever. Retriever is used for reported balance sheet and income statement items. To retrieve the data of transitional balance sheet items, as well future non- cancellable operating leasing fees, annual reports of 2018 or 2018/2019 for each company are used, complemented by annual or Q1 reports from 2019 in cases where sufficient information is missing in the annual reports of 2018 (2018/2019). Information of market capitalization, as of the year-end, is collected from the annual reports or from Nasdaq.

4.3 Statistical methods of hypothesis testing

To be able to properly measure the effects from the IFRS 16 transition, we make some important assumptions described in the following section. First, when calculating the change in total liabilities and assets (i.e. the total liabilities and assets post-implementation), we are using the transitional effects from the addition of right-of-use assets and lease liabilities reported by the

(20)

companies in the annual reports of 2018 (2018/2019). The reporting of transitional effects is similar, but not exactly uniform, among the firms in the sample. Some differences can exist between, for example, the reported right-of-use asset and the effect on total assets, following that prepayments and deferred tax assets etc. can also be affected by the IFRS 16 adoption (see examples in Appendix A). In many cases, companies provide information explicitly stating the effects on total assets and liabilities. In some cases, however, companies do not explicitly clarify if the reported right-of-use assets and lease liabilities differ from the effects on total assets and liabilities (Appendix A). If such differences exist, they are likely to be small, why the amount of the reported right-of-use assets and lease liabilities are in such cases treated as effects on total assets and liabilities. A small degree of uncertainty thereby exists, due to this lack of information. However, we assess this to be the most valid and reliable method of measuring the effects, since the alternative to using reported numbers would be to use a model of estimation for the full sample. By using the firms’ reported numbers, based on actual discount rates and similar assumptions, a contribution is made to the field of research by measuring more precise effects than prior studies (e.g. Fülbier et al., 2008; Durocher, 2008; Duke et al., 2009).

Second, since transitional numbers are only available for balance sheet items, another approach needs to be taken regarding EBITDA. The aim is to analyze the effects in 2019 (2019/2020) – using information available for the fiscal year ending at the date of transition – and the effect is assessed to be mainly the removal of operating lease expenses (Mulford & Gram, 2007). Some companies do, in the annual reports of 2018 (2018/2019), report their own estimations of the IFRS 16 effect on the upcoming year’s EBITDA. In these cases, the reported numbers are used in the estimation of EBITDA post-implementation. In most annual reports, however, estimated effects on EBITDA are not reported and due to this, we estimate the EBITDA post- implementation by removing the non-cancellable operating lease expenses, reported in the annual reports of 2018 (2018/2019), to be paid within the following year (see examples in Appendix A). Since the removal of these fees from the operating expenses are the main regulatory changes in EBITDA, inflicted by IFRS 16, this method of estimation is assessed to be very similar to the company’s own estimations in the cases where these are reported.

If a company would enter into a new lease contract after the publication of the annual report, this amount’s effect on EBITDA would of course not be covered by our estimation. However, since the numbers used are from the year-end closest to the IFRS 16 implementation, we assess these reported amounts to be the most valid proxies for estimating the effect. By making the

(21)

for the regulatorily mandated removal of non-cancellable operating leasing fees, we isolate the effect caused by IFRS 16.

Following the procedures described above, the effects (%) on total liabilities, total assets and EBITDA are calculated according to the following equations:

𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄 𝟏𝟏: 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑎𝑎𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑡𝑡𝑎𝑎 (%) =Total assets post implementation Total assets pre implementation − 1 𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄 𝟐𝟐: 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑎𝑎𝑡𝑡 𝑡𝑡𝑖𝑖𝑎𝑎𝑙𝑙𝑖𝑖𝑡𝑡𝑖𝑖𝑡𝑡𝑖𝑖𝑎𝑎𝑎𝑎 (%) =Total liabilities post implementation

Total liabilities pre implementation − 1 𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄 𝟑𝟑: 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑎𝑎 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 (%) =Estimated EBITDA post implementation

EBITDA pre implementation − 1

When calculating the change in the D/E ratio, potential transitional effects on equity also need to be regarded. Such effects exist due to the IFRS 16 standard enabling different methods of calculating the transitional balance sheet effects, of which some are making retroactive enforcement, causing transitional effects also to the opening balance of equity. In some cases, effects on equity can also follow from making provisions for uncertain tax positions. Because of these reasons, reported effects on the opening balance of equity is included in the calculation of changes in D/E:

𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄 𝟒𝟒: 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑎𝑎 𝐸𝐸/𝐸𝐸 (%) =Total liabilities post implementation / Equity post implementation Total liabilities pre implementation / Equity pre implementation −1 Enterprise value (EV) is commonly defined as Total liabilities + Market capitalization - Cash

& Cash equivalents (Loughran & Wellman, 2011). The effect on EV/EBITDA is calculated as following, in line with equation 1-4:

𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄 𝟓𝟓: 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑎𝑎 𝐸𝐸𝐸𝐸/𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 (%) =

(Total liabilities post implementation + Market capitalization − Cash & Cash Equivalents)/ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑖𝑖𝑖𝑖𝑝𝑝𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑝𝑝𝑖𝑖𝑝𝑝𝑖𝑖𝑝𝑝𝑖𝑖 (Total liabilities pre implementation + Market capitalization − Cash & Cash Equivalents)/ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑝𝑝𝑝𝑝𝑖𝑖 𝑖𝑖𝑖𝑖𝑝𝑝𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑝𝑝𝑖𝑖𝑝𝑝𝑖𝑖𝑝𝑝𝑖𝑖 −1

Using these equations, the effects are calculated as relative changes in percentages for each firm in the sample, in line Morales-Díaz & Ramírez-Zamora (2018a, 2018b). Using SPSS, the data is examined as a full sample and divided per sector, arriving at a mean and a median change for each type of change, along with minimum and maximum values etc. for each group.

(22)

In order to test the hypotheses of this study, statistical testing is needed. When concluding if a significant difference exists between two populations, such as before and after a change, the Student’s t-test is commonly used. However, this test requires the tested samples to be normally distributed, an assumption needed to be examined before performing any statistical testing.

Whether or not our full sample of percentage changes in total assets, total liabilities, EBITDA, EV/EBITDA and D/E is normally distributed is examined using the Test of Normality in SPSS (table 3). For a sample to be assessed as normally distributed, it needs to have a significance (p) value above 0.05 according to either the Kolmogorov-Smirnov or Shapiro-Wilk test (Pallant, 2016, p. 65), depending on sample size. Since our sample consist of 279 (>50) cases in each calculation, the Kolmogorov-Smirnov test is used for interpretation. To assess normality, skewness and kurtosis also need to be examined. According to Pallant (2016, p. 58) skewness and kurtosis values should be close to 0 if the data is normally distributed. If being larger than the absolute value of 1, the data is generally assessed not to meet the assumptions of a normal distribution.

Table 3: Tests of normality for the full sample of firms

Tests of Normality

Kolmogorov-Smirnov Shapiro-Wilk

Statistic df Sig. Statistic df Sig. Skewness Kurtosis Change in total assets % 0.299 279 0.000 0.462 279 0.000 5.476 39.286 Change in total liabilities % 0.305 279 0.000 0.437 279 0.000 6.801 64.57 Change in EBITDA 2018 % 0.396 279 0.000 0.202 279 0.000 11.348 147.586 Change in EV/EBITDA % 0.326 279 0.000 0.356 279 0.000 8.045 121.772

Change in D/E % 0.31 279 0.000 0.427 279 0.000 6.599 58.163

Note: The table consists test statistics for of all five measures analyzed in the study. It presents the test statistic, the degrees of freedom and the significance level of the tests of normality of the respective measure. In this study, it is only the Kolmogorov-Smirnov test of normality that is used for interpretation. The table also presents values of skewness and kurtosis.

From table 3, it is evident that our sample of percentage changes is clearly non-normally distributed. For all five measures, the Kolmogorov-Smirnov significance value is 0.000 (<0.05) and the sample is largely skewed (all measures >5) and has a high kurtosis value (all measures

>30). These findings are in line with our expectations, based on prior studies (Fülbier et al, 2008; Morales-Díaz & Ramírez-Zamora 2018a, 2018b) concluding that changes in total assets, liabilities etc. due to lease capitalization are not generally normally distributed. Following that our sample is skewed for all measures, this study mainly focuses on interpretation of the median values in our findings, rather than the mean changes. Median is a non-parametric statistic in

(23)

58). Differences between the two values do, however, tell something about the distribution in our sample and because of this, these differences are also discussed in chapter 5.

Since the sample is not normally distributed, non-parametric tests are needed to address our hypotheses. For the purpose of testing differences between pre- and post-implementation numbers (hypotheses 1 and 3), the Wilcoxon signed-rank test is used, in line with previous studies (Fülbier et al., 2008; Morales-Díaz & Zamora-Ramírez, 2018a, 2018b). This test is considered to be the non-parametric alternative to the repeated measures t-test, and it is designed to measure the same participants or data under two different conditions, i.e. using dependent samples (Pallant, 2016, p. 197). The Wilcoxon test converts scores to ranks and compares them at Time 1 and Time 2. In the case of this study, a sample of 279 pairs of pre- and post-implementation values in percentages is used. This test is used for determining if the change in total assets, total liabilities, EBITDA, EV/EBITDA and D/E is significantly different from zero (i.e. the pre-implementation period), in the full sample and for each sector.

Through the Wilcoxon test, a z-value is calculated. However, the z-value is based on the sum of signed ranks and not on the mean values of observed changes – as the repeated measures t- test is – hence, the Wilcoxon z-value only helps explain if the median change is significantly positive, negative, or not significant. Following our procedure of calculating differences in the sample (post-implementation minus pre-implementation), a significantly positive Wilcoxon z- value implies a median increase in the post-implementation data, and a significantly negative z-value implies a median decrease. Important assumptions for using the Wilcoxon signed-rank test are that the observations are paired and consist of the same participants or companies being re-tested on different occasions (i.e. dependent samples), that the pairs are independent, and that the data is continuous and measured on at least an interval scale (Pallant, 2016). Our data meet these assumptions. Hence, the Wilcoxon signed-rank test can be used for testing our first and third hypothesis.

For the purpose of testing sectoral differences in the post-implementation data (hypothesis 2), the Kruskal-Wallis test is used. The Kruskal-Wallis test is a non-parametric statistical test for comparing scores between three or more groups. Similar to the Wilcoxon signed-rank test, the Kruskal-Wallis test does this by converting the scores to ranks. By comparing the mean ranks of the different groups, the test is able to determine if the groups are significantly different from each other (Pallant, 2016, p. 199). The Kruskal-Wallis test statistics do not by themselves explain anything about which groups are significantly different (Pallant, 2016). However, SPSS

(24)

have a built-in function of post hoc-testing in the form of pairwise comparisons, available when performing a Kruskal-Wallis test. This test is used in this study (table 7 in chapter 5). Post-hoc testing of differences between two independent samples can also be performed in a separate function, using the Mann-Whitney u test (Pallant, 2016).

When performing multiple pairwise tests, like this study does, the alpha values need to be adjusted using the Bonferroni adjustment. The adjustment is made by dividing the alpha value (alternatively multiplying the p-value) with the number of pairwise tests performed (Pallant, 2016, p. 202). In the built-in pairwise comparison of the Kruskal-Wallis test samples that is used in this study, SPSS is automatically making Bonferroni adjustments to all p-values.

Following that the Kruskal-Wallis test is a non-parametric test, the test assumptions are similar to the Wilcoxon signed-rank test, however, requiring independent samples (e.g. post- implementation effects in different sectors) opposite to the Wilcoxon signed-rank test.

4.4 Methodological considerations

A common criticism of studies on effects from regulation is that it can be difficult to isolate the effect from the investigated variable when time is an issue. When observations are made in different time periods, other variables than the identified are likely to also affect the financial statement measures between period 1 and period 2. Hence, it is uncertain to what extent the measured change is an effect of the investigated variable. In this study, this problem is managed through analysis of the moment of transition, i.e. measuring the difference between for example 2018-12-31 and 2019-01-01. In normal circumstances, there are no differences between the closing balances of year 1 and the opening balances of year 2. Hence, when differences between the two periods are observed, the effect from IFRS 16 is isolated quite precisely.

Concerning the relative change in EBITDA, the above problem is more applicable. By using the moment of transition for collecting information on the effects on EBITDA in the upcoming year, as well as using the most recently reported EBITDA pre-implementation, we assess this problem being sufficiently managed. However, criticism could be expressed regarding that the EBITDA of the last year of IAS 17 might not be fully representative for the hypothetical EBITDA according to IAS 17 in the first year after the implementation of IFRS 16. It could also be criticized that new lease contracts could be entered or pro-longed during the year post- implementation. Hence, when using the non-cancellable contracts as of the transition date, this could underestimate the effect in the full year.

(25)

Following that our sample is quite small when divided into sectors, problems could occur in the significance testing. The first action taken into consideration is the reclassification of the four smallest sectors (described in section 4.2). Even so, the sample sizes of some sectors could still be criticized for being problematically small (<30) for performing significance testing (Pallant, 2016). However, this is mainly the case when performing parametric tests, e.g. the Student’s t- test. By using non-parametric tests, e.g. the Wilcoxon signed-rank test and the Kruskal-Wallis test, smaller sample sizes can be used due to that a normal distribution is not a required assumption.

(26)

5. Results

In the following sections, our findings regarding the effects from the IFRS 16 implementation in Swedish publicly listed firms are described and the three hypotheses are answered.

5.1 Results from the full sample

The full sample results, following the equations of change in total assets, total liabilities, EBITDA, EV/EBITDA and D/E described in the methods section, are presented in table 4 below. Presented are the mean and median values, together with standard deviation, minimum and maximum values, as well as the Wilcoxon z-values.

Table 4: Descriptive statistics of the full sample of firms

Measure Δ Tot. assets Δ Tot. liabilities Δ EBITDA Δ EV/EBITDA Δ D/E

N 279 279 279 279 279

Median 4.7% 9.5% 11.2% -6.4% 9.6%

Mean 10.1% 21.7% 39.6% -9.9% 22.3%

Wilcoxon z 14.478** 14.478** 14.478** -11.466** 14.478**

Min 0.1% 0.1% 0.1% -206.8% 0.1%

Max 190.5% 499.5% 2133.3% 587.6% 499.5%

Sd 19.1% 42.2% 149.6% 44.3% 44.6%

Notes: The table consists of statistics for all five measures analyzed in this study. It presents the number of cases (N), the median and mean values, the Z-value of the Wilcoxon signed-rank test, the minimum and maximum values observed, and the standard deviation (Sd).

* Significant on the 5% level

** Significant on the 1% level

5.1.1 Effects on financial statements

For the three measures of change in financial statements – total assets, total liabilities and EBITDA (table 4) – there are major effects following the implementation of IFRS 16. The median effect of relative increase is larger for total liabilities than for total assets but are both smaller in magnitude than the observed median increase in EBITDA. Further, all three measures show larger mean increases than median increases, a difference caused by the existence of high maximum values in all three measures. The maximum values, ranging between 190.5% and 2133.3% for the three measures, deviate much more extensively from the median values than the minimum values of 0.1% do. This indicates that, for a number of firms, the implementation of IFRS 16 did completely change the structure of the financial statements, more than doubling the size of the liabilities. These findings are interesting from the perspective of Duke et al.s (2009) argument of information previously being withheld from stakeholders, a topic further discussed in chapter 6. However, the hypothesis testing in this study is performed by

(27)

The differences between the relative increases in total assets and total liabilities are mainly attributable to the assets, pre-implementation, being larger than the liabilities. The actual amount of the new assets (mainly right-of-use assets) added through the IFRS 16 adoption were in most cases the exact same as the amount of the added liabilities (mainly lease liabilities), following that most companies have chosen the most simplified method of transition. A small number of companies in the sample did report small differences between added assets and liabilities, often due to them using methods of transition including retroactive calculations, causing some effects also on the opening balances of equity. The observed differences between the percentage change of total assets and liabilities are, however, still mainly attributable to the larger size of the pre-implementation assets.

The relative change in EBITDA is largely depending on the pre-implementation size of EBITDA. Following that some companies such as growth companies or companies in financial distress might report very low EBITDA in relation to expenses, such as from leases, the wide range of results is not surprising. Indeed, the maximum 5 values of relative changes in EBITDA all consist of companies reporting a positive or negative EBITDA close to zero. Even though a wide range is evident, the full sample consists only of positive changes (i.e. increases compared to pre-implementation). Hence, the findings are in line with our expectations.

Considering the results of the Wilcoxon signed-rank test, the median changes in total assets, total liabilities and EBITDA are all positive and significant on the 1% level. The reason behind the Wilcoxon z-values being the same in all three measures (table 4) is because it is based on ranks, and 14.478 is the largest possible z-value in a sample of 279 pairs of cases. This result follows from all cases showing the same sign (i.e. an increase). The results enable us to confirm our first hypothesis and conclude that the implementation of IFRS 16 caused a significant median increase in total liabilities, total assets and EBITDA of Swedish publicly listed firms.

5.1.2 Effects on key financial ratios

For the measures of change in key financial ratios, there is a negative median change in the EV/EBITDA multiple and a positive median change in the D/E ratio (table 4). As with the effects on the financial statement measures, there are wide ranges in the data and following this, there are distinct differences between mean and median values. The change in the D/E ratio is in all respects (median, mean and range) very similar to the change in total liabilities, following that equity is in most cases not affected on the day of transition to IFRS 16. As briefly discussed above, effects (mainly decreases) on equity do exist in a number of companies, following a

(28)

choice of a less simplified method of transition, and this is the cause of the small differences that do exist between changes in the D/E ratio and in total liabilities.

The observed median change in the EV/EBITDA multiple is negative, in contrast to the four other measures. As previously shown, the median relative increase in EBITDA is larger than the relative increase in total liabilities (affecting EV), which gives a first explanation to the negative change in the multiple. Further, since the nominator (EV) is affected not only by total liabilities, but also of market capitalization and cash and cash equivalents, the relative change of EV is smaller than the relative change of total liabilities. Since such a mitigating effect does not exist in the denominator (EBITDA), this further explains the observed decrease in the EV/EBITDA multiple. Another specific feature for the EV/EBITDA multiple is that the range of data consists of both positive and negative observations. However, the relation between the mean and median change is still in line with the other measures; the mean change being larger than the median change, which brings further clarity to the distribution of the sample and the variations in effects between firms. However, only the median change is used for hypothesis testing.

Considering the results of the Wilcoxon signed-rank test of the changes in the EV/EBITDA multiple and the D/E ratio, they are both significant on the 1% level. The change in EV/EBITDA has a negative z-value and the change in D/E has a positive z-value, thus confirming a significant median decrease in EV/EBITDA and a significant median increase in D/E. These results enable us to confirm our third hypothesis and conclude that the implementation of IFRS 16 caused significant median changes in the D/E ratio and EV/EBITDA multiple of Swedish publicly listed firms.

5.2 Results by sector

The results are presented by sector in table 5 below, following the same procedures as the calculations presented from the full sample in table 4. Presented are the mean and median values, minimum and maximum values, as well as the Wilcoxon z-value. Examining the different median values in the sectors gives an indication of an existence of sectoral differences, and these differences are statistically tested using the Kruskal-Wallis test (table 6). In table 5, the sectors are sorted by the size of the median change in total liabilities, sorted largest to smallest. The largest and the smallest observed median effects for all five measures are highlighted in bold in the table.

Implications of IFRS 16 adoption Evidence from Swedish publicly listed firms