Equity financial instruments: use, determinants and usefulness from IAS 39 to IFRS 9

From January 2018, International Financial Reporting Standard (IFRS) 9 Financial Instruments replaced International Accounting Standard (IAS) 39 Financial Instruments: Recognition and Measurement [International Accounting Standards Board (IASB), 2020b]. One of the major changes in this new standard was the classification and measurement requirements for investments in equity instruments (hereafter EFAs). IFRS 9 requires all fair value measurements for EFA and the default presentation of the fair value gains or losses (FVGL) on EFA in profit or loss (IFRS 9, para 4.1.2 and 5.7.5). However, the standard allows firms to present the FVGL on certain EFA in other comprehensive income (OCI) and prohibits recycling of the cumulative FVGL on these EFA upon their derecognition (International Accounting Standards Board (IASB), 2019, para 4.1.2 and 5.7.5). Considering the potential impact on net profit, Sir David Tweedie, the former International Accounting Standards Board (IASB) Chair, expressed concerns about the potential abuse of this choice (Street, 2014). The IASB constituents also hold divergent views on this choice, the prohibition on recycling and the usefulness of the FVGL on EFA presented in OCI (EFRAG, 2015, 2022). However, it is less clear whether it is being abused in practice.

IAS 39 uses an available-for-sale category for EFA and requires the FVGL on these assets to be presented in OCI (International Accounting Standards Board (IASB), 2009b, para 55). Further, the standard requires that FVGL “previously recognised in OCI shall be reclassified from equity to profit or loss as a reclassification adjustment” at the time of derecognition (IAS 39, para 55). Several studies observe that firms tend to sell EFA to boost reported profit when the recycling of the FVGL on EFA is allowed (Barth et al., 2017; Dong and Zhang, 2018; Lu et al., 2023). However, the choice to classify certain EFA at fair value through OCI (FVTOCI) is irrevocable and the recycling of the FVGL on these EFA is not permitted under IFRS 9, which makes the EFA FVTOCI classification a major decision in this setting. The IASB allowed the classification choice in IFRS 9 on the belief that the separate presentation of the FVGL on certain EFA in OCI would be useful to investors.

Despite the above changes in the classification and measurement of EFA under IFRS 9, we do not have much evidence on how these changes affect firms and the value relevance of EFA. While the literature on the effects of IFRS 9 is gradually growing, limited attention has been given specifically to the classification and measurement of EFA (Awuye and Taylor, 2024). Pinto and Morais (2022) and Fang et al. (2022) include some discussion on EFA classification in their studies, providing evidence from Europe and China, respectively. We contribute to this developing literature by providing empirical evidence from Australia on whether the standard affects changes in EFA usage, what factors determine the classification of EFA and how the standard impacts the value relevance of EFA.

We focus on Australia because it adopts IFRS word for word and has a very rigorous enforcement regime (Brown et al., 2014; Thomson, 2009). As a member jurisdiction that follows the IASB’s pronouncements, Australia adopted IFRS 9 in line with its global implementation timeline, ensuring consistency with international accounting practices. Further, the international accounting literature demonstrates that financial reporting is heavily shaped by the institutional setting in which it is embedded (Isidro and Raonic, 2012; Nobes, 1998; Soderstrom and Sun, 2007). Thus, the effects documented in this study are likely the results of the new standard rather than a lack of rigorous enforcement. Australia’s adherence to IFRS 9 facilitates cross-country comparability and enhances the potential generalisability of the findings to other IFRS-adopting, English-speaking common law jurisdictions with similar institutional environments.

Using data from the Australian Securities Exchange (ASX) 500 firms three years before and after IFRS 9 adoption, we first find that there is no significant change in the use of EFA, including whether to invest in EFA or not, EFA classification choice, or holding amounts, across both financial and non-financial firms after IFRS 9’s implementation. This result is consistent with findings in European banks (Löw and Erkelenz, 2022) but contrary to a Chinese study, which finds firms decrease their available-for-sale asset holdings before the mandatory adoption of the Chinese equivalent of IFRS 9 (Fang et al., 2022). We propose that the difference could perhaps stem from the variations in the notice period given before the mandatory application of IFRS 9. Specifically, the period from issuance to implementation of IFRS 9 is approximately four years, relative to less than one year in China. Prior literature documents that familiarity with the standard enhances the confidence of preparers and investors in accounting information (Alali and Foote, 2012; Mala and Chand, 2015). Thus, we contribute to the debate of whether to recycle FVGL on EFAs by showing that there is no impact on firms’ use of EFA in practice and suggest that a longer notice period before mandatory adopting IFRS 9 could facilitate the familiarity of the standard and mitigate firms’ potential costs from avoiding being adversely impacted by standard changes.

Secondly, we investigate the determinants of EFA classification choice in non-financial and financial firms both before and after IFRS 9, respectively. The results show that the EFA effect on net income is the main determinant for the choice of EFA classification in non-financial firms before IFRS 9. In contrast, the EFA amount, which could reflect the underlying economics of the investment, is significant post-IFRS 9. This suggests a decrease in the potentially opportunistic use of the available discretion of accounting for the FVGL from EFAs post-IFRS 9. We do not find that EFA characteristics are a key driver for financial firms either pre- or post-IFRS 9. Thus, our second contribution highlights the potential exploitation of the classification discretion given in IAS 39 and one potential benefit of IFRS 9 in constraining the opportunistic use of discretion, given the prohibition on recycling.

Finally, we investigate whether the EFA amount and FVGL on EFAs are value relevant. We find that the EFA amount is value relevant for financial firms or when it is large for non-financial firms, with no change in the value relevance pre- and post-IFRS 9. There is no consistent evidence that the EFA effects on OCI or profit or loss are more value relevant post-IFRS 9. However, Pinto and Morais (2022) find the EFA effect on OCI is value relevance post-IFRS 9 only after IFRS 9 in a sample of the top 100 UK and 50 European firms. Considering prior research also has mixed views on the usefulness of OCI components (Cahan et al., 2000; Isidro et al., 2004; Khan et al., 2018), our results are important as they highlight the need for caution in any optimistic interpretation of IFRS 9 in improving financial information usefulness.

These results would be of interest primarily to the IASB, national standard-setters, such as the Australian Accounting Standards Board and national accounting enforcement bodies, such as the Australian Securities and Investments Commission, who are interested in knowing how the recent standard changes affect financial statements and the usefulness of the resulting financial statement numbers. The result that the underlying economics of EFA, rather than opportunism, tend to drive the classification of EFA, would be of interest to investors and auditors who are interested in the existence of bias in the classification of EFA.

The rest of this paper is structured as follows. Section 2 provides the standard-setting background of equity instruments accounting, reviews the prior research literature and develops the hypotheses. Section 3 discusses the sample and research design. Section 4 reports on the use of EFA and descriptive statistics of the sample. Section 5 presents the results of the determinants and usefulness of the EFA classification choice. Section 6 concludes this study with a discussion.

The Global Financial Crisis brought to light many issues relating to the intricacy and opaqueness of accounting for financial instruments, prompting the IASB to replace the financial instruments accounting standard IAS 39 with IFRS 9 [International Accounting Standards Board (IASB), 2014]. A completed version of IFRS 9 was issued in July 2014 with an effective date of on or after 1st January 2018, with early application permitted ^[1].

One cause of financial instruments’ accounting complexity comes from the classification and measurement of EFA. Under IAS 39, EFA is classified as held for trading if it is acquired for the purpose of selling in the near term or if there is evidence of a recent actual pattern of short-term profit-taking (IAS 39, para 9). However, the default EFA classification is available-for-sale assets, which is a residual category that captures assets that do not meet the criteria of any of the other categories (e.g. held-to-maturity assets or loans and receivables) within the standard, and management has the intention to hold them for a longer period (BDO, 2018; Taylor, 2017). EFA shall be measured at fair value except for those that do not have a quoted price in an active market and whose fair value cannot be reliably measured, which shall be measured at cost [International Accounting Standards Board (IASB), 2008, para IN4 and BD2; IAS 39, para 43 and 46]. IFRS 9 removes the classification categorises and cost measurement from IAS 39 and results in the default classification of EFA as fair value through profit or loss (FVTPL) but allows fair value through other comprehensive income (FVTOCI) option (Barnoussi et al., 2020; IFRS 9, para 4.1.2 and 4.1.4). The IASB allows the FVTOCI option to address firms’ strategic investments, which are more of holding than trading instruments (Street, 2014).

Under IAS 39, the cumulative FVGL on available-for-sale EFAs that are “previously recognised in OCI shall be reclassified from equity to profit or loss as a reclassification adjustment upon derecognition (IAS 39, para 55)”. However, IFRS 9 prevents firms’ FVGL on EFAs from being recycled to net income when EFA is derecognised once FVTOCI is chosen at initial recognition (IFRS 9, para 5.7.5). The prohibition of recycling brought intense discussion when the IASB was developing IFRS 9. The European Financial Reporting Advisory Group (EFRAG) considers that the irrevocable FVTOCI option for EFA is unlikely to appeal to long-term investors, and decision usefulness may be reduced without recycling (EFRAG, 2015; Löw and Erkelenz, 2022). Sue Lloyd, the former IASB vice chair, said, “Recycling can provide a confusing presentation of performance. The Board’s view is that when an investment is held for strategic purposes (i.e. the intended narrow population), these gains and losses are not part of an investor’s performance (Lloyd, 2018)”. Thus, whether recycling should be allowed and the impact on decision usefulness remains an area of concern for standard setters [International Accounting Standards Board (IASB), 2022].

Under IAS 39, management’s intended holding period is important in EFA classification; however, it introduces ambiguity and increases the possibility of earnings management when recycling FVGL to net income is allowed (Barth et al., 2017; Dong and Zhang, 2018; Lu et al., 2023). The IASB has the intention to reduce management discretion and subjectivity in classifying EFA by making the FVTPL a default classification under IFRS 9, but allows irrevocable election of FVTOCI at initial recognition (Elnahass et al., 2018; Mechelli et al., 2020).

The determinants of accounting choices when selecting one accounting method over another have been discussed in extant literature; however, there is a lack of evidence on the determinants of firms’ EFA classification choice, particularly in the context of IFRS 9’s implementation (Da Costa et al., 2020; Israeli, 2015). Using a sample of the top 100 UK and 50 European firms (FTSE 100 and EURO STOXX 50 firms), Pinto and Morais (2022) identify leverage as the main factor driving the reclassification of available-for-sale assets to FVTOCI in the IFRS 9 transition year, while also examining the determinants of firms’ EFA holdings before and after IFRS 9. Therefore, it is of interest to know what drives firms’ EFA classification choice and whether it changed after IFRS 9.

Firstly, we identify the amount of EFA that may indicate how a firm manages such assets. Due to differing business natures, non-financial and financial firms may have different motivations for investing in equity instruments. Pecking order theory suggests that non-financial firms distribute their excess cash to shareholders or prioritise the excess cash as a source of internal financing rather than investing in securities for capital gain. However, many EFAs are viewed as strategic investments with the intention of establishing or maintaining a long-term operating relationship with the investee entity [International Accounting Standards Board (IASB), 2009a, para BC68]. For non-financial firms, when EFA amount comprises a larger portion of total assets, management is more likely to ensure the investment aligns with the firm’s overall strategic goals, and is, accordingly a strategic investment. Therefore, we predict that the EFA amount is positively related to the FVTOCI choice after IFRS 9, since an EFA investment need not be strategic to be classified as available-for-sale assets as per IAS 39.

Financial firms are normally deeply integrated into the capital market with a primary focus on investment activities. In line with the business objective in financial firms, EFA is mainly held for generating profit through trading and diversifying investment portfolio risk. Under IAS 39, EFA classification may need to be based on managers’ holding intention to either trade in a short-term or hold for a longer period for risk management, regardless of the amount ^[2]. Similarly, financial firms’ approach to managing EFAs may remain unchanged after IFRS 9 – whether they trade for profit-taking or hold for risk management – which is unrelated to the EFA amount. Therefore, we predict that there is no relationship between EFA amount and its classification choice in financial firms, regardless of IFRS 9 adoption. We state our first determinant hypothesis as follows:

Secondly, making the EFA classification choice by considering the EFA effect on net income or the level of fair value hierarchy may imply firms’ opportunistic use of the choice. Literature documents earnings management behaviour through realising FVGL on available-for-sale assets in both financial and non-financial firms in countries that allow the recycling of the FVGL on EFAs. Barth et al. (2017) and Dong and Zhang (2018) find that the US banks manage earnings by selectively trading available-for-sale assets to realise their FVGL from equity to profit or loss. Lu et al. (2023) provide evidence in Chinese non-financial firms that only when firms’ net income is positive or FVGL on available-for-sale assets is large enough to offset negative net income, firms smooth their earnings through realised gains and losses on available-for-sale assets ^[3].

On the one hand, FVTOCI option enables firms to avoid earnings volatility when the fair value of EFA fluctuates and manipulate earnings when recycling is allowed under IAS 39. On the other hand, FVTPL option provides a potential source of earnings that can be reflected in net income immediately. However, since EFA classification may be based on the purpose of holding such assets, EFA effect on net income may not be considered when making classification decisions. Therefore, we examine whether EFA classification choice is used opportunistically by considering EFA effect on net income and state our second determinant hypothesis as follows:

Finally, when there is no quoted price in an active market, Level 3 fair value hierarchy can be applied to EFA measurement. Song et al. (2010) find that banks recognise greater changes in assets measured at Level 3 fair value when they have lower earnings. Laux and Leuz (2010) document that banks reclassify assets measured at Level 1 fair value to Level 3 fair value during the Global Financial Crisis to avoid recognising the impairment losses. It is possible that FVTPL is preferred when EFA is measured at Level 3 fair value compared to Levels 1 and 2, as firms have more discretion over the fair value movements and are able to control earnings volatility. Therefore, we examine whether Level 3 fair value measurement for EFA drives firms’ classification choice and state our third determinant hypothesis as follows:

EFA amount accounts for an important portion of firms’ total assets, particularly in financial firms (Awuye and Taylor, 2024; Dong and Zhang, 2018; Lu et al., 2023). Khurana and Kim (2003) find that fair value provides a higher quality of information than historical costs for available-for-sale assets in bank holding companies. Since IFRS 9 eliminates the cost alternative permitted by IAS 39 and mandates all fair value measurements for EFA, it is of interest to know the extent to which EFA amounts provide incremental explanatory power to firms’ share prices and whether it is changed after IFRS 9. We examine the value relevance of the EFA amount and state our hypothesis as follows:

Research on examining the usefulness of FVGL on EFAs is limited, and there are no conclusive results on the value relevance of OCI in the literature. Cahan et al. (2000) and Isidro et al. (2004) do not find incremental value relevance of OCI. However, Khan et al. (2018) find that fair value movement of available-for-sale assets is one of the two components that drive the value relevance of OCI. EFRAG expresses concerns about the relevance of reported net income if FVGL on EFAs cannot be recycled to profit or loss once FVTOCI is chosen under IFRS 9 (EFRAG, 2015). However, a different view suggests that if recycling does not improve users’ access to information, it should be abandoned without introducing complexity to financial reporting (Rees and Shane, 2012). Pinto and Morais (2022) examine a sample of FTSE 100 and EURO STOXX 50 firms and document that FVTPL option for EFA provides incremental value relevance both before and after IFRS 9, while FVTOCI option becomes value relevant after IFRS 9. Therefore, we examine whether the FVGL on EFAs is value relevant and whether the reporting location (either profit or loss or OCI) makes a difference to investors, especially when recycling is prohibited under IFRS 9, and state our hypothesis as follows:

Our initial sample comprises ASX 500 firms ^[4]. ASX 500 firms represent more than 90% of the market capitalisation of ASX firms, and firms that are not in the ASX 500 have few EFAs. Our main purpose is to examine the use, determinants and usefulness of EFA and whether they have changed after IFRS 9. Based on this objective, we require a firm to be listed on ASX three years before and after IFRS 9 adoption. The time-series data provides comparability and allows us to analyse the impact of IFRS 9 on firms’ EFA applications. We exclude 110 firms that do not have a six-year listing period (three years before and after IFRS 9 adoption) on ASX. We exclude managed funds, firms that do not use IFRS and insurance companies that are exempt from adopting IFRS 9 until IFRS 17 is effective. Panel A in Table 1 displays our sample selection process and results in a final sample of 2,262 observations with 377 unique firms.

We collect EFA information from firms’ annual reports by hand. Firstly, we identify the year of initial application, i.e. the year in which firms first apply IFRS 9. Since the IASB issued IFRS 9 in 2014 and allowed early adoption, 44 unique firms (11.7%), including 12 from the financial sector, applied for IFRS 9 earlier than its effective date ^[5]. The remaining 333 firms (88.3%) applied IFRS 9 after it went into effect. Given that the majority of our sample firms have 30th June as the balance date, 2019 is the most common year of initial application for ASX 500 firms. We denote the year of initial application as year t for firm i, and year t−1, t−2, t−3 and t+1, t+2 represent years before and after IFRS 9 adoption.

Secondly, we search firms’ annual reports to check whether they have EFAs or not. We use available-for-sale assets information under IAS 39 if there is no separate disclosure of the portion of EFA in available-for-sale assets. The term EFA is used both before and after IFRS 9 for simplicity. Panel B in Table 1 displays our sample distribution by Global Industry Classification Standard (GICS) sectors. In total, 39.48% of the sample (893 observations) have EFAs. Out of all the sectors, the financials have the highest percentage (80.65%) of observations (300 observations) that have EFAs, which is in line with the nature of the sector.

Thirdly, if the firm has EFAs, we then collect the balance amount, the classification choice (either FVTOCI or FVTPL), the FVGL effect and fair value hierarchy for firms’ EFAs. For firms that have multiple EFAs and disclose them in separate lines, we collect the aforementioned information for each EFA line item and aggregate it. If there is no disclosure of the FVGL on EFAs, we assume that it is either zero or immaterial and assign zero to FVGL for that firm-year. An example of how EFA information is disclosed in a firm’s annual report during the year of initial application can be found in  Appendix 1. All other firm data (e.g. firm financials, corporate governance, etc.) are collected either manually from firms’ annual reports or from the Refinitiv database.

A firm may have multiple equity investments that are classified into different categories. We define a firm as an FVTOCI user if any of its EFA is classified as FVTOCI ^[6]. We specify the following logistic model to test H1 over samples that have EFAs:

where FVTOCI is an indicator variable that equals one if a firm i is an FVTOCI user in year t; zero otherwise. EFAamt is the EFA balance amount at year-end deflated by total assets in testing H1a. To test H1b, EFAeffect measures EFA effect on net income and is calculated as the absolute value of a ratio of any FVGL on EFAs to net income. For H1c, EFAmeas is an indicator variable that equals one if firm i uses Level 3 fair value hierarchy to measure any of its EFAs in year t, and zero otherwise.

We control for firm contracting incentives since prior literature provides evidence that accounting choice is determined to influence one or more contractual arrangements (Alves, 2019; Fields et al., 2001; Murphy, 2000). Firstly, existing research documents that firms with high financial leverage are at greater risk of breaching debt covenants and are more likely to choose the accounting method to reduce earnings volatility (Israeli, 2015). Pinto and Morais (2022) find that firms with high leverage have greater contractual risk and reclassify more available-for-sale assets to FVTOCI only during the IFRS 9 transition year. We include LEV, which is the leverage calculated by total liabilities divided by total assets, as a control. Secondly, literature shows that managers take advantage of the discretion allowed in accounting standards to increase their compensation, especially short-term incentives (Guidry et al., 1999; Healy, 1985; Murphy, 2000). EFA classification choice affects firms’ net income, which has an immediate effect on profitability. As an important metric of business performance, net income is a common factor in considering management compensation (De Angelis and Grinstein, 2015; Graham et al., 2005). We control for CEOCOMP, which is computed as the ratio of the CEO’s variable compensation (including cash bonus and equity awards) to total compensation, in our model.

Following Pinto and Morais (2022), we also control for firm size (LogTA) and profitability (ROEadj). Larger firms are more likely to have more surplus cash and invest in EFAs. Therefore, we control for firm size and measure it as the natural logarithm of total assets (LogTA). Da Costa et al. (2020) and Barlev et al. (2007) suggest that firms with a high return on assets (ROA) would face greater demand from others and are likely to choose income-increasing accounting methods. We use return on equity before any FVGL on EFAs effect (ROEadj) rather than ROA, since ROE is more consistent with the investor decision-making objective of financial reporting (Zang et al., 2022) ^[7].

Moreover, we control for corporate governance factors that monitor firms’ compliance with accounting standards for financial reporting quality improvement (Schäuble, 2019). We include audit committee independence (ACIND) as a control since effective monitoring by the audit committee improves the accuracy of financial estimates and constrains opportunistic actions made by managers (Ashbaugh-Skaife et al., 2006; Dechow et al., 1996; Larcker et al., 2007; Srinidhi et al., 2011). We control for BIG4 if firm i is audited by the Big 4 in year t. Literature documents that prestigious auditors are more likely to identify managers’ opportunistic behaviour and have a strong incentive to enforce higher earnings quality (Francis and Wang, 2008). We also control for analyst coverage (ANALYST) because a high analyst coverage contributes to firms’ information disclosure and monitors firms’ financial reporting quality (Yu, 2008).

To test whether the determinants of firms’ EFA classification choices have been changed after IFRS 9, we examine the determinants model in samples before and after IFRS 9 separately. With different holding purposes, we estimate the model for non-financial and financial firms separately. To control for potential industry variations in EFA classification choice as well as time-specific effects, we include industry and year fixed effects ^[8]. All continuous variables in this study are winsorised at one per cent on both tails to minimise outliers’ influence. Standard errors are clustered at the firm level. The detailed definitions of variables are listed in  Appendix 2.

To extend our knowledge regarding the relevance and reliability of equity instruments’ accounting as reflected in equity values, we test the value relevance of (1) EFA amount and (2) FVGL on EFAs. Based on an extensively used Ohlson (1995) model, we follow other value relevance studies and estimate the price-level regression model over samples that have EFAs (Barth et al., 1996, 2001; Khan et al., 2018; Liao et al., 2021):

where Price = the share price of firm i in year t, three months after its balance date; BVE = book value of equity; CI = total comprehensive income. All variables except Price are deflated by the number of outstanding shares (denoted by _S). Although many studies use net income in the value relevance model (Barth et al., 1996; Ciftci et al., 2014; Pinto and Morais, 2022), we use CI, because the Ohlson (1995) model is based on the clean surplus rule. Further, considering the objective of examining the usefulness of FVGL on EFAs in both profit or loss and OCI, comprehensive income fits in this study.

To investigate the incremental value relevance of EFA amount, we isolate the EFA amount (AmtEFA) from BVE and include AmtEFA as a separate component of the model. We also separate FVGL on EFAs in profit or loss or/and OCI from CI to examine the value relevance of FVGL. The effect of EFA on profit or loss is denoted as PLEFA and on OCI is denoted as OCIEFA. If there is no disclosure of firms’ EFA effect, we assume it is zero and immaterial:

where BVE_S* is computed as BVE_S – AmtEFA_S. CI_S** is CI_S excluding PLEFA_S and OCIEFA_S. To examine whether value relevance changed after the adoption of IFRS 9, we test the model on samples before and after IFRS 9 for non-financial and financial firms separately.

To address our first research objective of understanding the use of EFA and whether it changed after IFRS 9, we examine firms’ EFA holding behaviour from three perspectives: whether to invest in EFA or not, EFA classification choice and EFA amount. Figure 1 presents the percentage of firms that have EFAs and the use of FVTOCI classification throughout the six years surrounding IFRS 9 adoption in non-financial and financial firms, respectively.

As shown in Panel A of Figure 1, the percentage of non-financial firms that have EFAs is around 31% over the six years around IFRS 9 adoption, with no significant variation in the Chi-square test (untabulated). In line with Zang et al. (2022), we also do not find significant changes in the percentage of FVTOCI users of firms that have EFAs with the Chi-square test, even though there is a decrease from 68.0% one year before IFRS 9 to 58% in the initial adoption year. Table 2 displays the descriptive statistics of EFA amount to total assets. The EFA amount comprises around 3% of total assets on average in each of the six years around IFRS 9 adoption. We compare EFA amount one year before and after the adoption of IFRS 9 with a t-test and do not find significant differences.

Panel B of Figure 1 shows EFA use and classification choice in financial firms. Using the Chi-square test, there is no significant difference in whether or not to invest in EFA over the six years, given that around 80% of financial firms have EFAs each year. About 52% of financial firms with EFAs are FVTOCI users each year, with no significant changes over the six years. In each of the six years around IFRS 9 adoption, the EFA amount represents, on average, about 37.7% of the total assets of financial firms, as shown in Table 2. Again, we do not find a significant difference in EFA amount one year before and after IFRS 9 adoption in financial firms with a t-test.

The results show that IFRS 9 does not change the use of EFA in both non-financial and financial firms regarding whether or not to invest in EFA, EFA classification choice, or EFA amounts, and this result is consistent with Zang et al.’s (2022) study in Australia and Löw and Erkelenz (2022)’s study in European banks. However, Fang et al.’s (2022) study in China finds that firms sell their available-for-sale assets after the announcement but before the implementation of the Chinese equivalent of IFRS 9 to avoid being adversely affected by new accounting standards. We note that the notice period given before the mandatory adoption of IFRS 9 is different between China and Australia. The IASB completed the requirements for changing EFA accounting standards under IFRS 9 in July 2014 and required mandatory adoption almost four years later. However, China announced its Chinese version of IFRS 9 in March 2017 and required implementation of the same as IFRS 9 in January 2018 (Fang et al., 2022). Literature documents that familiarity with the standards facilitates greater confidence in preparers and investors in the quality of accounting information (Alali and Foote, 2012; Liu and Liu, 2007; Mala and Chand, 2015). The short notice period of less than one year in China may cause firms to overreact to the financial consequences and respond by selling available-for-sale assets. The results imply that a longer notice period before the mandatory adoption of IFRS 9 facilitates familiarity with the standards and mitigates the potential cost caused by overreaction to the standard changes. The change in the recycling requirement for FVGL on EFAs from equity to profit or loss upon derecognition has no practical effect on EFA use.

Panel A of Table 3 presents summary statistics of variables used in the determinants model of equation (1) in non-financial and financial firms before and after IFRS 9 application, respectively. On average, the EFA amount (EFAamt) comprises 3.0% of total assets with a median value of 0.7% in non-financial firms before IFRS 9 and 3.0% (0.5%) on average (median) after IFRS 9. The average (median) EFA amount is 37.3% (13.0%) of total assets in financial firms before IFRS 9 and 38.2% (14.7%) after IFRS 9. The mean (median) FVGL on EFAs (EFAeffect) comprises 8.3% (0.3%) of net income before IFRS 9 and 8.5% (0.2%) after IFRS 9 in non-financial firms. The FVGL on EFAs accounts for a higher portion of net income in financial compared to non-financial firms, as reflected in the average EFAeffect of 35.6% before and 36.5% after IFRS 9. Firms that have a Level 3 fair value hierarchy to measure any of their EFAs (EFAmeas) increased after IFRS 9 in both non-financial (76 firms increased to 103 firms) and financial firms (39 firms increased to 53 firms).

Panel B of Table 3 contains descriptive statistics for variables used in the value relevance model. The average (median) share price three months after its balance date (Price) is 6.3 (2.6) before and 8.0 (3.0) after IFRS 9 in non-financial firms, compared to 11.9 (5.1) before and 12.9 (4.3) after IFRS 9 in financial firms. The mean value of EFA amount deflated by total outstanding shares (AmtEFA_S) is 0.16 before and 0.19 after IFRS 9 in non-financial firms, compared to a larger amount of EFA in financial firms with an average AmtEFA_S of 3.1 before and 3.0 after IFRS 9. The average OCIEFA_S for non-financial firms is 0.001 before and −0.007 after IFRS 9, respectively, whereas for financial firms, it is 0.025 before and 0.008 after IFRS 9, respectively. In non-financial firms, the average PLEFA_S is 0.000 before and 0.002 after IFRS 9, whereas in financial firms, it is 0.013 before and 0.012 after IFRS 9. Detailed variable definitions can be found in  Appendix 2.

Panel A in Table 4 shows the results of EFA classification determinants in non-financial firms before and after IFRS 9, respectively. As shown in Column (1), non-financial firms are more likely to choose FVTOCI classification when EFA effect has a lower impact on net income before IFRS 9, given the coefficient −0.795 for EFAeffect. The result is consistent after we control for firm contracting incentives, firm size, profitability and corporate governance, showing that the negative relationship between EFAeffect and FVTOCI choice is significant at the 1% level, as displayed in Column (3). Column (2) presents the results of EFA classification determinants in firm-years after IFRS 9, showing EFAamt is significantly positively associated with FVTOCI choice. After including all the control variables, we find a similar result that the FVTOCI choice is preferred when EFA amount is large after IFRS 9, as shown in Column (4).

The results imply that there is a possible opportunistic use of the discretion given in IAS 39 by non-financial firms considering EFA effect on net income when making the classification decision. Non-financial firms may prefer to reflect FVGL on EFAs immediately in earnings when the effect is greater. In contrast, under IFRS 9, a larger EFA amount might indicate a strategic investment, for which FVTOCI is chosen in alignment with the standard.

Next, Panel B in Table 4 presents the results of the determinants model in financial firms. As shown in Columns (1) and (2), EFA characteristics are not related to FVTOCI choice, regardless of IFRS 9 adoption. After controlling for firm contracting incentives, firm size, profitability and corporate governance, as shown in Columns (3) and (4), we find consistent results that EFA amount, EFA effect on net income or EFA measurement are not related to FVTOCI choice.

Since the initial year of IFRS 9 adoption is the only year that firms can reclassify their existing EFAs, we examine the determinants model on a sub-sample of firms in year t, as shown in Column (5), in both non-financial and financial firms. We find consistent results.

Overall, the determinants of EFA classification choice in non-financial firms changed from an EFA effect on net income before IFRS 9 to the EFA amount after IFRS 9. We infer that the potential opportunistic use of the EFA classification discretion in standards is mitigated after IFRS 9, although there does not appear to be a change for financial firms.

Considering different EFA investing purposes, we test the value relevance model in non-financial and financial firms separately in both pre- and post-IFRS 9 periods, as shown in Table 5. Firstly, we report the baseline model that book value of equity and total comprehensive income are significantly positively associated with share price in both non-financial and financial firms before and after IFRS 9, as shown in Columns (1)–(2) and (5)–(6) in Table 5, and the results are consistent with existing literature (Khan et al., 2018; Rees and Shane, 2012).

Secondly, we separate the EFA amount from the book value of equity and the effect of FVGL on EFAs from comprehensive income as reported in Columns (3)–(4) and (7)–(8) before and after IFRS 9 in non-financial and financial firms, respectively. As shown in Columns (3)–(4), both EFA amount and FVGL on EFAs do not provide incremental value relevance for non-financial firms, regardless of IFRS 9 adoption. In contrast, the EFA amount is value relevant for financial firms both before and after IFRS 9 [Columns (7)–(8)]. OCIEFA_S is significantly positively associated with share price in financial firms before but not after IFRS 9, implying that IFRS 9 decreases the usefulness of EFA effect on OCI for financial firms.

Again, we conduct a number of robustness tests. Firstly, we further examine the value relevance model in a sub-sample of firm-years whose EFA amount is larger than the median. EFA amount becomes value relevant in both non-financial and financial firms after IFRS 9, suggesting IFRS 9 improves the value relevance of EFA amount when it is material to firms’ assets. Secondly, we examine the value relevance model in a sub-sample of larger firms. Thirdly, the value relevance model is tested in sub-samples that have a larger EFA effect, either in profit or loss or in OCI. All sub-sample tests consistently show that OCIEFA_S provides value relevance only before IFRS 9 for financial firms.

Pinto and Morais (2022) find that EFA effect on OCI becomes value relevant after, but not before, IFRS 9 in the top 100 UK and 50 European firms; however, their value relevance models do not include any variable capturing the EFA amount, nor do they conduct separate analyses for financial and non-financial firms. In contrast, we find that the amount of the EFA is value relevant both pre- and post-IFRS 9 when material in both financial and non-financial firms. These differences could be the result of variations in institutional settings, sample size, constituents and models. Accordingly, our results suggest caution in the interpretation of IFRS 9 in improving information usefulness by having the market value of EFA amounts.

Accounting for financial instruments has long been an area of concern, especially after the Global Financial Crisis (Duh et al., 2012; PwC, 2017). Due to its complexity, a completed and integrated standard for financial instruments-IFRS 9 – was not issued until 2014 as a replacement for IAS 39. IFRS 9 was effective on or after 2018, with early application permitted. One major change in IFRS 9 regarding the classification and measurement of EFA is the prohibition of recycling FVGL on EFAs from equity to profit or loss once FVTOCI is chosen at initial recognition. The contentious issue of whether to recycle FVGL on EFAs has been debated widely.

The findings from a sample based on firms in the ASX 500 three years before and after IFRS 9 adoption show that the use of EFA in practice, in terms of whether to invest in EFA or not, EFA classification choice and holding amounts, did not change significantly in both non-financial and financial firms after IFRS 9. There is a potential opportunistic use of the EFA classification discretion before IFRS 9 because the EFA effect on net income was an important driver of recognising the FVGL in OCI or the profit or loss in non-financial firms. However, this appears constrained after IFRS 9, where the amount of EFA is significant instead. Finally, we find that although the amount of EFA is value relevant, there is no change post-IFRS 9.

Our paper is among the first to contribute to the effect of IFRS 9 implementation. It has several implications for the IASB, national standard-setters and national accounting enforcement bodies:

• a longer notice period before the mandatory application of IFRS 9 allows stakeholders, including managers, investors, preparers and auditors, more time to familiarise themselves with the requirements of the new standard and facilitates a smooth transfer from IAS 39 to IFRS 9;

• there appears to be less use of the EFA classification choice to impact earnings under IFRS 9; and

• standard setters need to be cautious when evaluating the information improvement in IFRS 9.

This study is subject to three limitations, which provide potential opportunities for future research. Firstly, this study examines the use, determinants and usefulness of changes in accounting standards for EFAs, but it does not address the potential costs associated with those changes, such as the impact on audit fees, labour costs or the cost of capital. Future research could extend the analysis by examining the costs related to the EFA standard changes. Secondly, our study focuses on the Australian setting. Since the international accounting literature documents that the impacts of accounting standards may vary depending on the institutional settings (Isidro and Raonic, 2012; Soderstrom and Sun, 2007), future studies may investigate the effects of the changes in EFA accounting standard in other settings. This is important because IFRSs are applied globally; therefore, the IASB would be interested in evidence on the impacts of IFRS 9 from different settings to form an overall opinion on how the standard affects financial reporting. Thirdly, while our research design and empirical models mitigate the potential omitted variable bias, the paper may still suffer from this bias.

This study is based on research conducted for Zeting Zang’s PhD thesis, “Accounting for Equity Financial Instruments under International Financial Reporting Standard (IFRS) 9 Financial Instruments: Use, Determinants, Usefulness, and Cost”, completed at Auckland University of Technology, 2025. The authors would like to thank the editor, Charl de Villiers, and the anonymous reviewers for their constructive comments. They also appreciate the valuable comments provided by Daifei (Troy) Yao, Warrick van Zyl, Asheq Rahman, and Borhan Bhuiyan on early drafts.

Conflict of interest statement: There is no material conflict of interest.

Table A1 

Figure A1