Timeliness of the Financial Reporting Among Maltese Licensed Voluntary Organisations

Governed by the G​o​v​e​r​n​m​e​n​t​ ​o​f​ ​M​a​l​t​a​ ​(​2​0​0​7​), the voluntary sector in Malta comprises over 1,800 registered voluntary organisations and 30,000 volunteers (T​i​m​e​s​ ​o​f​ ​M​a​l​t​a​,​ ​2​0​2​4). The Government of Malta recognises the sector as a vital socio-economic pillar in society (F​a​r​r​u​g​i​a​ ​P​o​r​t​e​l​l​i​,​ ​2​0​2​4). A comparative study on volunteering across the European Union highlights the heterogeneity of the voluntary sector, with organisations linked to political and cultural aspects, religious initiatives, social welfare activities, and sports. Religious organisations and those involved in social work activities together represent around 90% of the sector (N​G​E​n​v​i​r​o​n​m​e​n​t​,​ ​2​0​2​1). Additionally, a strong culture of volunteering has been a well-established tradition for around a century (G​H​K​,​ ​2​0​1​0), with nearly all NGOs being managed by volunteers (N​G​E​n​v​i​r​o​n​m​e​n​t​,​ ​2​0​2​0). In a recent survey, Malta actually ranks in the top ten, with an index score of 54 points, which evidences the local culture’s disposition to help those in need (C​h​a​r​i​t​i​e​s​ ​A​i​d​ ​F​o​u​n​d​a​t​i​o​n​,​ ​2​0​2​4).

In the non-profit sector, maintaining accountability and stakeholders’ trust is crucial (B​r​a​n​d​t​n​e​r​,​ ​2​0​2​1; K​a​b​a​,​ ​2​0​2​1; O​r​t​e​g​a​-​R​o​d​r​í​g​u​e​z​ ​e​t​ ​a​l​.​,​ ​2​0​2​0; T​a​c​o​n​ ​e​t​ ​a​l​.​,​ ​2​0​1​7) as this can either enhance or inhibit the organisations’ ability to operate effectively. Sustaining stakeholder trust is deemed to be a critical factor for their legitimacy and effectiveness (D​e​t​h​i​e​r​ ​e​t​ ​a​l​.​,​ ​2​0​2​3; G​i​b​e​l​m​a​n​ ​&​ ​G​e​l​m​a​n​,​ ​2​0​0​1; H​a​r​r​i​s​ ​e​t​ ​a​l​.​,​ ​2​0​1​7; J​e​a​v​o​n​s​,​ ​2​0​0​1) and is also theorised to be a fundamental condition for charitable giving, as argued by G​a​s​k​i​n​ ​(​1​9​9​9​) and C​h​a​p​m​a​n​ ​e​t​ ​a​l​.​ ​(​2​0​2​1​). Two key arguments underpin this line of reasoning. First, individuals with a stronger general disposition to trust others are considered more inclined to engage in charitable giving compared to those with lower levels of trust. Empirical evidence supports this claim. For instance, B​e​k​k​e​r​s​ ​(​2​0​0​3​) found a positive relationship between generalised trust and the level of charitable donations in the Netherlands. B​r​o​w​n​ ​&​ ​F​e​r​r​i​s​ ​(​2​0​0​7​) found such a positive relationship for the USA, and similar claims have been found across other countries, (E​v​e​r​s​ ​&​ ​G​e​s​t​h​u​i​z​e​n​,​ ​2​0​1​1; G​l​a​n​v​i​l​l​e​ ​e​t​ ​a​l​.​,​ ​2​0​1​6). A second line of reasoning emphasises that non-profit organisations, as fundraisers for specific causes or beneficiary groups, rely on trust through the mechanism of perceived efficacy, that is, the expectation that charitable contributions will meaningfully benefit the recipients (B​e​k​k​e​r​s​ ​&​ ​W​i​e​p​k​i​n​g​,​ ​2​0​1​1). Empirical evidence, such as that of B​e​l​d​a​d​ ​e​t​ ​a​l​.​ ​(​2​0​1​5​) and H​o​u​ ​e​t​ ​a​l​.​ ​(​2​0​1​7​), supports this claim: the former study found that American donors’ belief in the efficacy of their contribution predicts their willingness to continue donating, while the latter found that donor satisfaction exerts a strong influence on future behaviour and motivation for donating.

Financial reporting is a mechanism used to enhance accountability, transparency and trust, as it enables stakeholders to assess the financial position of the entity and how its funds were obtained and utilised. In 2018, the National Audit Office in Malta warned that the Office of the Commissioner for Voluntary Organisations (OCVO), which is the regulatory authority responsible for this sector, was unable to ascertain how much money was being raised by voluntary organisations (VOs) or how it was being used (N​A​O​,​ ​2​0​1​8​a). The report further highlighted a notably high rate of non-compliance among VOs regarding the timely submission of the annual return. From a user’s perspective, prior research, such as B​l​a​n​c​h​e​t​ ​(​2​0​0​2​), H​e​a​l​d​ ​(​2​0​0​6​), and K​u​l​z​i​c​k​ ​(​2​0​0​4​), and, views timely reporting as an important concept in transparency, alongside other key concepts including accuracy, consistency, appropriateness, completeness, and clarity. In financial reporting, timeliness is “having information available to decision makers before it loses its capacity to influence decisions” (R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4). Delays in financial reporting may therefore diminish the relevance of the disclosed information, potentially suggesting weaknesses in governance or operational effectiveness, and undermining the confidence of stakeholders (A​l​-​A​j​m​i​,​ ​2​0​0​8; K​n​e​c​h​e​l​ ​&​ ​P​a​y​n​e​,​ ​2​0​0​1). This may be detrimental to non-profit organisations. “Timeliness thus plays an important role in this regard” (A​d​e​b​a​y​o​ ​&​ ​A​d​e​b​i​y​i​,​ ​2​0​1​6).

Accordingly, regulators acknowledge the importance of timely reporting by imposing deadlines for reporting. D​a​o​u​d​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) describe the financial reporting lag (FRL) as “the period between the end of the financial reporting period and the date the financial reports are issued, or the date of the submission of the reports to the regulatory bodies.” The FRL in this study refers to the length of time between the end of the reporting period of a VO and the date when the annual return and accounts of the VO are submitted to the OCVO.

R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) note that an understanding of the determinants of the FRL in the voluntary sector is important. However, literature in connection with the FRL in the voluntary sector is quite limited. Most studies on the FRL focus on publicly traded firms, typically examining their influence on market reactions or identifying firm-specific determinants such as size, audit complexity, or sectoral characteristics (C​h​a​m​b​e​r​s​ ​&​ ​P​e​n​m​a​n​,​ ​1​9​8​4; L​e​v​e​n​t​i​s​ ​&​ ​W​e​e​t​m​a​n​,​ ​2​0​0​4). In contrast, limited research addresses timeliness in non-profit organisations, despite their reliance on public funds, grants, and tax advantages (K​e​a​t​i​n​g​ ​e​t​ ​a​l​.​,​ ​2​0​0​5). VO’s financial statements are an important source of information for various stakeholders, such as governments, donors, and volunteers. As argued by Y​e​t​m​a​n​ ​&​ ​Y​e​t​m​a​n​ ​(​2​0​1​1​), inaccurate financial statements can lead to suboptimal decisions and the potential misallocation of resources. This financial dependence intensifies the need for timely reporting to ensure accountability and preserve public trust. As such, examining the FRL in the Maltese non-profit context offers valuable insights into transparency and governance practices in a sector where stakeholder scrutiny is high and resources are scarce. In light of this, the present study:

(a) Measures the extent of the FRL in Category 2 and Category 3 VOs during the years 2018–2020; and

(b) Ascertains the important explanatory variables affecting the FRL.

This article is intended to contribute to the literature on financial reporting in the voluntary sector. It is particularly timely in view of the focus on non-profit organisations and the launch, in October 2025, of INPAS, the first internationally recognised financial reporting standard for the non-profit sector. It is expected that the adoption of this standard by non-profit organisations will strengthen the rigour of their financial reporting processes and reinforce stakeholders’ confidence in the integrity of the reported financial information (S​a​i​d​y​k​h​a​n​ ​&​ ​S​e​n​s​o​y​,​ ​2​0​2​3). The sector is also currently being eyed for reform by the Maltese Government and the regulator to strengthen it, enabling it to continue being recognised as an important pillar in society (M​i​n​i​s​t​r​y​ ​f​o​r​ ​I​n​c​l​u​s​i​o​n​ ​&​ ​t​h​e​ ​V​o​l​u​n​t​a​r​y​ ​S​e​c​t​o​r​,​ ​2​0​2​4). A public consultation document was issued in 2024. Regulatory frameworks shape organisations’ conduct and their transparency practices and are therefore important (D​e​t​h​i​e​r​ ​e​t​ ​a​l​.​,​ ​2​0​2​3). Additionally, while VOs in Malta are deemed to be “a main pillar within society and the economy” (O​C​V​O​,​ ​2​0​2​5), there is a paucity of scholarly research. The C​e​n​t​r​e​ ​f​o​r​ ​E​u​r​o​p​e​a​n​ ​V​o​l​u​n​t​e​e​r​i​n​g​ ​(​2​0​2​0​) noted that the minimal “research on volunteering in Malta significantly limits the degree to which national and organisational policies and services aimed at promoting and supporting volunteering can ever be developed in an appropriate, effective, efficient manner.” In this connection, this study provides the regulator with evidence-based research to inform future policy changes regarding non-profit organisations in Malta.

The remainder of this paper is structured as follows. The next section reviews the relevant literature and develops the study hypotheses. Section 3 outlines the data collection and the methodology employed to examine the explanatory variables affecting the financial reporting timeliness of VOs. Section 4 presents the findings, and Section 5 discusses the results, while the study concludes with Section 6.

Prior to 2007, Malta’s voluntary sector lacked formal regulation, with no legal obligations in the areas of registration, financial reporting, or operational transparency. The 2005 White Paper, “Strengthening the Voluntary Sector”, introduced formal regulatory oversight. In 2007, the Voluntary Organisations Act was enacted, through Act XXII of 2007, establishing a regulatory framework and the OCVO, to initiate oversight over the sector while ensuring transparency and accountability.

Under the Voluntary Organisations Act, a VO is defined as an organisation established for any social purpose, operating on a non-profit basis, and comprising volunteers. Registration of such organisations with the OCVO involves meeting several requirements, including the submission of annual returns and accounts, as introduced by Legal Notice 379 of 2012 (G​o​v​e​r​n​m​e​n​t​ ​o​f​ ​M​a​l​t​a​,​ ​2​0​1​2). VOs are classified into three categories by size, based on the annual revenue and income of the organisation. The size thresholds over the years, including the changes brought about in 2020 through Legal Notice 317 of 2020 (G​o​v​e​r​n​m​e​n​t​ ​o​f​ ​M​a​l​t​a​,​ ​2​0​2​0) are depicted in Table 1.

In 2013, the OCVO reported that only 55% of VOs submitted their respective annual returns and accounts within the deadline stipulated by law. A year later, the OCVO reported that the percentage of submissions across all categories was decreasing over the years. At this point, the concept of the FRL came under the scrutiny of the National Audit Office (NAO). Through a performance audit, the NAO concluded that for the financial year ending 2016, only 16% of VOs were compliant in submitting their annual returns and accounts within the legal timeframe, with most non-compliance relating to Category 1 VOs, owing to a lack of technical and administrative competencies (N​A​O​,​ ​2​0​1​8​b). To address this, key legislative changes were enacted in 2018: the OCVO’s responsibilities were expanded, the thresholds were revised as depicted in Table 1, mandatory enrollment for all VOs was required, and enforcement actions in cases of non-compliance were addressed. Additionally, prior to 2018, Category 2 VOs could choose to prepare their accounts on either a cash basis or an accruals basis, and this had to be clearly disclosed in the accounts. One of the legal amendments in 2018 required Category 2 VOs to prepare their accounts in line with GAPSME, hence adhering to the accruals concept. At the same time, the OCVO launched several initiatives to enhance its internal processes, empowering the Commissioner to reinforce its regulatory role in the vetting and oversight of voluntary organisations (O​C​V​O​,​ ​2​0​1​9). For VOs, “the ability to fundraise depends entirely on retaining the public’s trust” (M​a​l​t​a​ ​C​o​u​n​c​i​l​ ​f​o​r​ ​t​h​e​ ​V​o​l​u​n​t​a​r​y​ ​S​e​c​t​o​r​,​ ​2​0​2​1). Following the aforementioned legislative changes, a Gallup and Wellcome Trust study found that Malta ranked among the top ten countries globally, with 79% expressing trust in charities and only 11% expressing distrust (Y​o​u​n​i​s​ ​&​ ​R​z​e​p​a​,​ ​2​0​1​9). Notably, no other EU countries made the global top ten (T​h​e​ ​M​a​l​t​a​ ​I​n​d​e​p​e​n​d​e​n​t​,​ ​2​0​1​9).

The returns submitted by VOs are currently regulated by the Voluntary Organisations Act and the VOs (Annual Returns and Annual Accounts) Regulations enacted through Legal Notice 317 of 2020. Legal Notice 317 repealed Legal Notice 379 of 2012 and assigned different deadlines for VOs to submit annual returns and annual accounts depending on their category, as detailed in Table 2. Additionally, the Commissioner’s powers in cases of non-compliance were further strengthened. The Commissioner may now request any information from any person to assess whether an organisation qualifies as a non-profit entity and whether the principles, rules, or guidelines of the law are being observed. Furthermore, Article 34 of the Voluntary Organisations Act provides that, upon investigation, the Commissioner can also impose penalties for breaches of the law’s provisions, as well as request information from banks or other entities.

On the other hand, VOs enrolled with the OCVO may enjoy the privileges contemplated in the law. Under article 4 of the Voluntary Organisation Act, these include being able to make collections without the need to obtain further authorisations; also receiving or being the beneficiary of grants, sponsorships or other financial aid from the Government, any entity controlled by the Government or the Voluntary Organisations Fund. These changes in legislation stem from the importance of VOs maintaining accountability towards stakeholders, reflecting both upward accountability to funders and regulators, and downward accountability to beneficiaries, employees, and volunteers, such accountability being on the lines indicated by the literature (D​h​a​n​a​n​i​ ​&​ ​C​o​n​n​o​l​l​y​,​ ​2​0​1​2; O​’​D​w​y​e​r​ ​&​ ​U​n​e​r​m​a​n​,​ ​2​0​0​7; U​n​e​r​m​a​n​ ​&​ ​O​’​D​w​y​e​r​,​ ​2​0​0​6). A study carried out in 2024 through surveys and focus groups with businesses and VOs in Malta found that acquiring detailed data and encouraging transparent reporting are essential for a more informed and equitable giving sector in Malta (A​c​a​d​e​m​y​ ​o​f​ ​G​i​v​e​r​s​,​ ​2​0​2​4). It is clear, therefore, that good reporting is important in building trust (C​o​n​n​o​l​l​y​ ​&​ ​H​y​n​d​m​a​n​,​ ​2​0​1​3​a), as is timely reporting. However, while financial reporting in the voluntary sector aims to enhance transparency and accountability (Y​a​s​m​i​n​ ​&​ ​G​h​a​f​r​a​n​,​ ​2​0​2​1) and is helpful to stakeholders (C​o​n​n​o​l​l​y​ ​e​t​ ​a​l​.​,​ ​2​0​1​7; T​r​u​s​s​e​l​ ​&​ ​P​a​r​s​o​n​s​,​ ​2​0​0​7), little empirical work has been undertaken to investigate the timeliness aspect of financial reporting in the voluntary sector. In one of the earliest studies focusing on the FRL in the voluntary sector, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) report that, in the Belgian voluntary sector, around 17.2% of VOs within the selected sample filed their annual financial statements late in 2007 and 2008. Similarly, C​h​r​i​s​t​i​a​e​n​s​ ​e​t​ ​a​l​.​ ​(​2​0​1​5​) found that for the years 2011 and 2012, among large and very large VOs in Belgium, 24% of the sample failed to meet the legal deadline. A similar percentage was also found to apply in the case of registered charities in England and Wales, with 18% filing their financial statements late (M​o​r​g​a​n​,​ ​2​0​1​1). More recently, M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​), drawing on a sample of UK charities from 2007 to 2018, reported a relatively low proportion of late filings at 5.02%, indicating that most charitable organisations file their financial statements on time. In fact, in the UK, the filing rates amongst UK charities have improved with 92% of charities filing on time by the end of 2024–25, representing a significant improvement from the previous year’s 81% (C​h​a​r​i​t​y​ ​C​o​m​m​i​s​s​i​o​n​ ​f​o​r​ ​E​n​g​l​a​n​d​ ​&​ ​W​a​l​e​s​,​ ​2​0​2​5).

This study focuses on the timeliness of financial reporting of VOs in Malta. As noted earlier, Legal Notice 317 of 2020 revised the date for filing annual accounts and other documents with the OVCO, as illustrated in Table 2. The option to prepare the accounts using a higher-level framework was also introduced at this time. Such documents are available to the public for inspection, allowing the regulator to conduct a proper assessment. The aim is to ensure transparency and accountability. The annual returns that are to be submitted to the OCVO include the (i) organisation chart; (ii) list of administrators; (iii) any documents amending the statute; (iv) annual return and annual report which require an overview of the finances during the year including the sources of funding, the fundraising activities if any, the disbursement of funds and the financial challenges that the VO has encountered; and a signed copy of the annual accounts prepared as outlined in Table 2. This illustrates additional obligations arising from Subsidiary Legislation 492.02 in relation to the assurance and filing deadline based on Legal Notice 317 of 2020.

The M​a​l​t​a​ ​C​o​u​n​c​i​l​ ​f​o​r​ ​t​h​e​ ​V​o​l​u​n​t​a​r​y​ ​S​e​c​t​o​r​ ​(​2​0​2​1​) noted that “not all voluntary organisations have the same capacity and thus there should not be any equal administrative burden which in principle also reflects the three categories established by the same Voluntary Organisations Act with reference to financial capacity”. Subsidiary Legislation 281.05—Accountancy Profession (General Accounting Principles for Small and Medium-sized Entities) Regulations (GAPSME) is the national GAAP of Malta, which is highly aligned with the IFRS Accounting Standards (M​i​c​a​l​l​e​f​,​ ​2​0​1​7). This was the result of the transposition of the EU Accounting Directive 2013/34/EU into Maltese legislation (M​i​c​a​l​l​e​f​ ​&​ ​G​a​u​c​i​,​ ​2​0​2​3), whose aims are to facilitate cross-border investment and enhance comparability and public confidence in financial statements through consistent disclosure. In terms of reporting requirements, minor and medium-sized for-profit organisations in Malta now need to prepare their financial statements on an accrual basis in line with GAPSME, unless the Board of Directors passes a resolution to the effect that IFRSs as adopted by the EU are to be used. On their part, large for-profit organisations and public-interest entities need to follow IFRSs as adopted by the EU. Interestingly, the thresholds in Table 1 are much lower than the financial reporting size thresholds in the Companies Act of G​o​v​e​r​n​m​e​n​t​ ​o​f​ ​M​a​l​t​a​ ​(​1​9​9​5​) (based on revenue applicable to for-profit organisations. This means that, for financial reporting purposes, Category 1 VOs are dissimilar to their for-profit counterparts of the same size, as the former use the cash basis and only file an Income and Expenditure template, thereby being subject to relatively fewer reporting requirements than their for-profit counterparts. Arguments supporting this could relate to the problems faced by these entities in terms of compliance costs, such as the lack of skilled accounting staff (C​o​r​d​e​r​y​,​ ​2​0​1​3) and high staff turnover (M​o​h​d​ ​N​a​s​i​r​ ​e​t​ ​a​l​.​,​ ​2​0​0​9). On the other hand, some Category 3 VOs are required to prepare IFRS financial statements when similarly sized for-profit organisations are required to abide by GAPSME. While IFRS Accounting Standards enhance the extent of comparability in financial reporting across entities, thereby promoting consistency and transparency in financial reporting (O​n​a​h​ ​&​ ​E​d​e​h​,​ ​2​0​2​4), the different thresholds of VOs relative to for-profit organisations mean that there is currently no level playing field. The current requirements might be imposing too much of a relative burden on these VOs, and this could ultimately materialise in a lack of financial reporting timeliness.

As of 2024, all for-profit organisations, with minor exceptions, were legally required to undergo audits. As Table 2 illustrates, Category 1 and Category 2 VOs are exempt from such stringent measures. However, in 2025, changes to the audit obligations of Maltese companies took effect through the Audit Exemption Rules, 2025 (G​o​v​e​r​n​m​e​n​t​ ​o​f​ ​M​a​l​t​a​,​ ​2​0​2​5). Such rules now mainly apply for financial periods commencing on or after 1 January 2025 and cater for new criteria under which companies may be exempt from statutory audit but are subject instead to a review engagement. Such companies include those qualifying as micro entities (in terms of turnover, if this is equal to or less than €93,000). Contrastingly, prior to this revision in law and throughout the period during which the fieldwork for this study was conducted, Category 1 and Category 2 VOs were not required to undergo an audit.

In 2024, OCVO introduced a new set of proposals for reforming the voluntary sector. The M​i​n​i​s​t​r​y​ ​f​o​r​ ​I​n​c​l​u​s​i​o​n​ ​&​ ​t​h​e​ ​V​o​l​u​n​t​a​r​y​ ​S​e​c​t​o​r​ ​(​2​0​2​4​) notes that the need to update the way the sector is regulated has long been felt, and such reform should provide the voluntary sector with the opportunity to continue its work. Among the proposals being put forward in this consultation process, termed VO Plus, are changes to the thresholds and the respective reporting procedures as illustrated in Table 3.

As depicted in Table 3, the 2024 consultation document proposed nearly doubling the thresholds across the three categories, thereby allowing more VOs to benefit from simpler financial reporting requirements. In December 2024, the Government of Malta published a feedback report on this consultation document. Such feedback showed that most VLOs are in favour. It was also suggested that “more research on the sector should be carried out” (M​i​n​i​s​t​r​y​ ​f​o​r​ ​I​n​c​l​u​s​i​o​n​ ​&​ ​t​h​e​ ​V​o​l​u​n​t​a​r​y​ ​S​e​c​t​o​r​,​ ​2​0​2​4). This study aims to contribute to the research process by shedding light on the FRL prevalent among VOs in Malta. In addition, as the usefulness of financial information is further evidently affected by several underlying factors contributing to the FRL of Maltese VLOs, the study also aims to shed light on the main explanatory variables that affect such FRL. To ensure sufficient comparability between the VOs within the scope of this study, it is limited to the accruals-based Categories 2 and 3 VOs, thereby excluding the cash-based Category 1 VOs. Furthermore, previous studies investigating the timeliness aspect of financial reporting in the voluntary sector, such as M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) in the UK charities and R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) in Belgium, have used regression analysis to identify determinants of financial reporting timeliness in these sectors. This study is therefore aimed at contributing to the non-profit literature by complementing these findings as applicable to VLOs in a small island state and also utilising neural networks rather than regression models. The principal advantage of this approach over traditional regression methods is that neural networks are considered more effective in modelling complex, non-linear relationships within data, given that regression analysis typically relies on the assumption of linearity and homogeneity of variance.

VOs are commonly classified into two main types: donative and commercial (C​h​a​n​g​ ​&​ ​T​u​c​k​m​a​n​,​ ​1​9​9​4; H​e​r​m​a​n​ ​&​ ​R​e​n​z​,​ ​2​0​0​8; O​’​H​a​g​a​n​ ​&​ ​P​u​r​d​y​,​ ​1​9​9​3). The key distinction between them lies in their revenue structure, with commercial VOs relying more extensively on non-donative income sources compared to their donative counterparts (C​a​l​a​b​r​e​s​e​,​ ​2​0​1​1). Resource dependence theory, proposed by S​a​l​a​n​c​i​k​ ​(​1​9​7​8​), posits that organisations rely on external stakeholders for essential resources, thereby rendering them susceptible to varying degrees of external influence. In the non-profit context, such critical resources tend to be donations, which makes the group of donors an important external stakeholder (C​o​n​n​o​l​l​y​ ​&​ ​H​y​n​d​m​a​n​,​ ​2​0​1​3​a; C​o​n​n​o​l​l​y​ ​&​ ​H​y​n​d​m​a​n​,​ ​2​0​1​3​b; H​y​n​d​m​a​n​ ​&​ ​M​c​D​o​n​n​e​l​l​,​ ​2​0​0​9). C​o​o​l​e​y​ ​&​ ​R​o​n​ ​(​2​0​0​2​) note that most VOs prioritise their accountability towards donors because their survival is dependent on them. Consequently, consistent with resource dependence theory, donative VOs seek to maintain compliance with legal and regulatory requirements, including the timely submission of reports (V​e​r​b​r​u​g​g​e​n​ ​e​t​ ​a​l​.​,​ ​2​0​1​1). Based on a survey carried out by the N​a​t​i​o​n​a​l​ ​S​t​a​t​i​s​t​i​c​s​ ​O​f​f​i​c​e​ ​(​2​0​0​8​) in Malta, a total income of €37 million was reported by the NGOs, and donations accounted for 44% of this total income. Donors are interested in the use of donated funds and therefore exert a degree of monitoring, often using the financial statements to evaluate how responsibly resources are used (C​o​n​n​o​l​l​y​ ​&​ ​H​y​n​d​m​a​n​,​ ​2​0​1​3​b; T​r​u​s​s​e​l​ ​&​ ​P​a​r​s​o​n​s​,​ ​2​0​0​7) to assess accountability. The signalling theory, initially developed by S​p​e​n​c​e​ ​(​1​9​7​3​), is also relevant due to the information asymmetry that exists between the non-profit organisations and the donors (H​a​r​r​i​s​ ​&​ ​N​e​e​l​y​,​ ​2​0​2​1). Donors continue to donate funds to the more accountable non-profit organisations (F​e​l​i​x​ ​e​t​ ​a​l​.​,​ ​2​0​1​7; H​y​n​d​m​a​n​ ​&​ ​J​o​n​e​s​,​ ​2​0​1​1; H​y​n​d​m​a​n​ ​&​ ​M​c​K​i​l​l​o​p​,​ ​2​0​1​8). Financial accountability would be reflected in good quality, transparent financial statements (H​y​n​d​m​a​n​ ​&​ ​M​c​D​o​n​n​e​l​l​,​ ​2​0​0​9; O​r​t​e​g​a​-​R​o​d​r​í​g​u​e​z​ ​e​t​ ​a​l​.​,​ ​2​0​2​0), which is based on the premise that signalling is employed to mitigate information asymmetries, and H​a​r​r​i​s​ ​&​ ​N​e​e​l​y​ ​(​2​0​2​1​) argue that transparency serves as a signalling mechanism aimed at building credibility and fostering a sense of connection and trust. Consequently, non-profit organisations are incentivised to decrease their FRL and to enhance their reporting so that donors are more willing to provide funds to the entity (H​y​n​d​m​a​n​ ​&​ ​M​c​M​a​h​o​n​,​ ​2​0​1​1; R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4).

Recent evidence suggests that organisations that rely on donations are more likely to submit their financial reports on time. One example is M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​), who examined 8,490 UK charitable companies from 2007 to 2018 and found that charities relying more heavily on donation income took a shorter time to file their financial reports. Another example is R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​), who found that in Belgium, VOs that are predominantly reliant on donations and/or grants have a greater incentive to disclose their financial results quickly. On the contrary, C​h​r​i​s​t​i​a​e​n​s​ ​e​t​ ​a​l​.​ ​(​2​0​1​5​) found that the resource dependency theory does not hold for “large” and “very large” VOs in Belgium, and no negative relationship was found between reliance on donations and/or grants and the FRL. Given that VOs in the local context are smaller in size compared to those in Belgium, and that “if VOs have capacity issues when it comes to reporting, that could stop Businesses from collaborating” (A​c​a​d​e​m​y​ ​o​f​ ​G​i​v​e​r​s​,​ ​2​0​2​4), the following hypothesis is formulated:

H1: There is a negative relationship between reliance on donations and/or grants and the FRL.

Previous research based on for-profit organisations, relating to the size of the organisation and the timeliness of financial reporting, shows opposing views. On one hand, some studies find an inverse relationship between organisation size and the time period, suggesting that larger organisations report in a more timely manner (A​d​e​b​a​y​o​ ​&​ ​A​d​e​b​i​y​i​,​ ​2​0​1​6; A​t​i​a​s​e​ ​e​t​ ​a​l​.​,​ ​1​9​8​9; E​z​a​t​ ​&​ ​E​l​‐​M​a​s​r​y​,​ ​2​0​0​8; G​i​v​o​l​y​ ​&​ ​P​a​l​m​o​n​,​ ​1​9​8​2; K​u​ ​I​s​m​a​i​l​ ​&​ ​C​h​a​n​d​l​e​r​,​ ​2​0​0​4; O​w​u​s​u​-​A​n​s​a​h​,​ ​2​0​0​0). This is often attributed to increased external monitoring (N​e​w​t​o​n​ ​&​ ​A​s​h​t​o​n​,​ ​1​9​8​9) and higher internal resources, such as accounting staff and systems (O​w​u​s​u​-​A​n​s​a​h​,​ ​2​0​0​0). Furthermore, since large organisations are more visible when compared to smaller ones, they are more inclined to adopt strategies aimed at mitigating the risk of regulatory intervention (A​l​-​A​j​m​i​,​ ​2​0​0​8). These arguments, in relation to private organisations, are likely to apply to non-profit organisations as well (R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4). However, in relation to VOs, taking income as the primary determinant of size, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) found a significantly positive correlation between size and the FRL, finding support in the argument put forward by B​r​y​a​n​t​-​K​u​t​c​h​e​r​ ​e​t​ ​a​l​.​ ​(​2​0​0​7​), who argued that the higher complexity of large organisations may delay the reporting process. These results were also supported by C​h​r​i​s​t​i​a​e​n​s​ ​e​t​ ​a​l​.​ ​(​2​0​1​5​), who found that when assets are used as a proxy for size, there is no negative relationship between the size of the VO and the FRL. Moreover, in Malaysia, Z​a​i​n​o​n​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) also found a positive relationship between size and the FRL; however, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) found that when assets were taken as the primary determinant of size, a negative relationship emerged between the size of the VO and the FRL. This can be attributed to the fact that measuring size in the voluntary sector is not straightforward (B​e​a​t​t​i​e​ ​e​t​ ​a​l​.​,​ ​2​0​0​1).

In Malta, VOs are frequently expected to operate in a manner akin to for-profit organisations, with requirements to develop business plans and articulate long-term strategic visions; however, given that volunteers and minimal staff essentially carry out their activities, the imposition of extensive regulatory and administrative demands poses significant challenges to their capacity for effective delivery (A​c​a​d​e​m​y​ ​o​f​ ​G​i​v​e​r​s​,​ ​2​0​2​4). Most NGOs operate with the help of voluntary workers (N​a​t​i​o​n​a​l​ ​S​t​a​t​i​s​t​i​c​s​ ​O​f​f​i​c​e​,​ ​2​0​0​8), and VOs often find themselves lacking the necessary tools and human resources. Moreover, when support is offered in the form of pro bono services, VOs acknowledge that they are often deprioritised due to the absence of payment, which subsequently creates challenges in meeting deadlines (A​c​a​d​e​m​y​ ​o​f​ ​G​i​v​e​r​s​,​ ​2​0​2​4). Consequently, based on this research, the following hypotheses are formed:

H2a: There is a positive relationship between the size of the VOs, measured by total revenue and income, and the FRL.

H2b: There is a positive relationship between the size of the VOs, measured by total assets, and the FRL.

Prior research posits that management is more likely to release favourable news quickly compared to bad, unfavourable news (B​a​s​u​,​ ​1​9​9​7; B​o​w​e​n​ ​e​t​ ​a​l​.​,​ ​1​9​9​2; B​r​y​a​n​t​-​K​u​t​c​h​e​r​ ​e​t​ ​a​l​.​,​ ​2​0​0​7; C​h​a​m​b​e​r​s​ ​&​ ​P​e​n​m​a​n​,​ ​1​9​8​4; H​a​w​ ​e​t​ ​a​l​.​,​ ​2​0​0​0). This is because good news may attract potential investors and help in retaining existing ones. Based on this theory, unfavourable financial information is associated negatively with the timeliness of financial reports and hence leads to a longer FRL. F​r​e​e​m​a​n​ ​(​2​0​1​0​) argues that organisations are accountable to various stakeholders who can influence or are influenced by the organisation. The origin of stakeholder theory dates back to the 1960s, when the Stanford Research Institute first proposed that organisations need the support of both their shareholders and stakeholders to survive (M​a​h​a​j​a​n​ ​e​t​ ​a​l​.​,​ ​2​0​2​3). In the non-profit context, stakeholders typically include donors, the government and regulators, beneficiaries, employees, volunteers, organisational partners, and the wider community (H​y​n​d​m​a​n​ ​&​ ​M​c​C​o​n​v​i​l​l​e​,​ ​2​0​1​8; V​a​n​ ​P​u​y​v​e​l​d​e​ ​e​t​ ​a​l​.​,​ ​2​0​1​2). According to stakeholder theory, such organisations need to address the information needs of multiple stakeholders to preserve legitimacy, accountability, and trust. Consequently, the acquisition and retention of contracts, grants, and other financial support depend on non-profit organisations demonstrating accountability through audited financial statements and various forms of performance reporting (L​e​r​o​u​x​,​ ​2​0​0​9).

In financial reporting, unfavourable financial information can take several forms, and some researchers examine various related hypotheses (R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4; T​r​u​s​s​e​l​ ​&​ ​P​a​r​s​o​n​s​,​ ​2​0​0​7). Organisations tend to delay filing their financial statements when reporting a deficit, negative equity, low liquidity and high leverage. To capture financial distress, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) test these variables in a non-profit context, and K​e​a​t​i​n​g​ ​e​t​ ​a​l​.​ ​(​2​0​0​5​) rely on these variables to assess their financial vulnerability. More recently, M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) found that charities with unfavourable news in their financial statements, represented by deficits, negative equity, low liquidity, and high leverage, tend to lag in their filings.

Although charities are not primarily driven by profit, reporting financial deficits may still be perceived negatively, similar to losses in the for-profit sector (H​y​n​d​m​a​n​ ​&​ ​M​c​C​o​n​v​i​l​l​e​,​ ​2​0​1​8; R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4). Furthermore, maintaining a modest financial surplus is commonly considered sound financial management within the non-profit sector (A​u​s​t​r​a​l​i​a​n​ ​C​h​a​r​i​t​i​e​s​ ​&​ ​N​o​t​-​f​o​r​-​p​r​o​f​i​t​s​ ​C​o​m​m​i​s​s​i​o​n​,​ ​2​0​2​2; J​i​m​e​n​o​ ​G​a​r​c​í​a​ ​e​t​ ​a​l​.​,​ ​2​0​2​3). Such surpluses contribute to the organisation’s financial sustainability, allowing it to meet both planned commitments and unforeseen costs, thereby enhancing long-term stability and public trust (A​n​h​e​i​e​r​,​ ​2​0​1​4; P​a​r​k​ ​e​t​ ​a​l​.​,​ ​2​0​2​2). In the Belgian voluntary sector, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) found a negative relationship between profitability and the FRL.

R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) also found a negative relationship between liquidity and the FRL in the Belgian voluntary sector, as well as a positive relationship between gearing and the FRL. C​a​l​a​b​r​e​s​e​ ​&​ ​G​r​i​z​z​l​e​ ​(​2​0​1​2​) found that debt may signal to donors that a non-profit organisation is at risk of insolvency, potentially leading to a reduction in donations. High levels of debt may also deter donations from donors who prefer their donations to be allocated toward current, rather than past, program activities (Y​e​t​m​a​n​,​ ​2​0​0​7). Additionally, donors prefer to fund non-profit organisations that are more established (P​a​r​s​o​n​s​,​ ​2​0​0​3), and they may become apprehensive as increasing gearing is deemed a sign of financial vulnerability.

Similar to for-profit organisations, a positive equity, with assets in excess of obligations, needs to be maintained in order to operate (A​n​t​h​o​n​y​,​ ​1​9​8​3). Without an adequate reserve of funds, a non-profit organisation will be unable to continue operating normally. It may become financially distressed when faced with a reduction in revenues, including donations, or in the event of financial problems (T​r​u​s​s​e​l​ ​&​ ​P​a​r​s​o​n​s​,​ ​2​0​0​7). T​r​u​s​s​e​l​ ​&​ ​G​r​e​e​n​l​e​e​ ​(​2​0​0​4​) find that equity is a positive and significant predictor of financial stability. Non-profit organisations with higher equity ratios can leverage their assets without curtailing their program activities. Consequently, organisations with greater equity are less likely to experience financial distress. R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) found a negative relationship between equity and the FRL in the Belgian voluntary sector. Based on this literature, the following hypotheses are formed:

H3a: The distribution of FRL differs across firms with negative versus positive profitability, with negative profitability expected to be associated with longer FRL.

H3b: VOs with low liquidity are expected to exhibit longer FRL compared with VOs with high liquidity.

H3c: VOs with low equity are expected to have longer FRL compared with VOs with high equity.

H3d: VOs with high gearing are expected to have longer FRL compared with VOs with low gearing.

Section 2.1 provides details on the legal framework for VOs in Malta, with one key amendment, implemented in 2018, primarily affecting the basis for preparing Category 2 VOs. Prior to 2018, Category 2 VOs could choose to prepare their accounts on either a cash basis or an accruals basis, and this had to be clearly disclosed in the accounts (L.N. 379 of 2012). This was changed in 2018 through the Voluntary Organisations (Amendment) Act (G​o​v​e​r​n​m​e​n​t​ ​o​f​ ​M​a​l​t​a​,​ ​2​0​1​8) requiring them to adhere to the accruals concept. Additionally, in 2018, the thresholds were raised across all three categories, resulting in a shift between categories and changes to the financial reporting adherence requirements. O​w​u​s​u​-​A​n​s​a​h​ ​(​2​0​0​0​) suggests that compliance improves over time as familiarity with new regulatory requirements increases, and L​u​y​p​a​e​r​t​ ​e​t​ ​a​l​.​ ​(​2​0​1​6​) found that, consistent with a learning curve effect, older small firms in Belgium are significantly more likely to file their financial statements early. Drawing from this literature, following the local amendment in 2018, a learning effect is expected, as administrators become increasingly familiar with the new procedures. Therefore, the following hypothesis is formed:

H4: The distribution of FRL differs across the years 2018–2020, with FRL expected to decrease over time.

To ensure comparability of financial reporting, this study focuses on Category 2 and Category 3 VOs, as these follow the accrual basis of accounting. As noted earlier, Category 2 follows the accrual basis of accounting since 2018, and Category 1 VOs prepare accounts on a cash basis. The population for this study, therefore, consists of all Category 2 and Category 3 VOs registered with the OCVO as of the end of 2017. This cut-off point was selected to minimise the likelihood of missing annual returns and accounts for the study period, which spans 2018–2020. Based on these criteria and discussions with the OCVO, the population for this study was determined to be 206 VOs. It is essential to note that while the VOs make submissions of the accounts to the OCVO at the end of each financial year, these are not accessible online on the OCVO’s website, but can be obtained through a formal request to the OCVO. Any required accounts for all the years 2018–2020 needed to be specifically identified and set aside by the staff at OCVO. Given the practical constraints of setting aside all the required accounts, the OCVO agreed to select a sample of 103 VOs through simple random sampling for each of the three years, resulting in 309 accounts being sourced. This sampling method minimises both recognised and unrecognised sources of human bias (T​h​o​m​p​s​o​n​,​ ​2​0​1​2). At a 95% confidence level, the study yields a margin of error of 6.84%. The accounts of these VOs for each of the three years were subsequently made accessible by the OCVO for analysis. All required data for the variables outlined in Section 3.2.1 were then manually extracted from these accounts, except for the variable FRL, which was extracted from a file provided by the OCVO containing the submission dates for every set of accounts.

This study examines the primary explanatory variables that influence the FRL within the Maltese voluntary sector. Prior to the neural network analysis, we conducted univariate analyses to explore the relationships between the predictor variables and the outcome. These analyses provided an initial assessment of associations. A neural network was then applied to capture complex, non-linear relationships among variables, to predict the outcome, and to determine which variables are more important than others. Results from univariate tests and neural networks may differ: a variable may not show a significant association in, for example, the Spearman or Kruskal-Wallis tests, but still contribute substantially to predictive performance in neural networks as the latter is comparing the dependent variable to the other explanatory variables, and additionally it determines which ones are significant and which are not significant. Furthermore, the application of neural networks in accounting research remains relatively novel, particularly within studies focused on the voluntary sector and the FRL.

Non-parametric statistical tests were conducted to assess the hypothesised relationships between various financial and organisational proxies and the FRL. Table 4 presents a description of the variables employed in this study together with the corresponding non-parametric test applied to each hypothesis. In addition, for data simplification the variables ‘FRL’; ‘Total Revenue and Income’; and ‘Total Assets’ were expressed as a logarithmic function. The variables’ description is mostly based on R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​). The results for each hypothesis are discussed in turn.

The neural network is a collection of nodes organised in a number layers as shown in Figure 1. Each node in the input layer represents one of the explanatory variables in that set. The number of nodes in the hidden layer is chosen automatically by SPSS; the number is optimised to produce the best possible results. The output layer contains one node that generates the predicted classification. The connections between the nodes, commonly termed edges, are typically assigned weights that indicate the strength of the relationship, with larger weights reflecting stronger links between two connected neurons. The output layer can therefore be modelled by a non-linear function involving the sum of the neurons in the input layer.

We denote the nodes in the input layer by $x_i$, where $i \in\{0,1,2,3, \ldots, P\}$. Here, P denotes the number of variables in the data set and $x_0$ denotes the so-called bias term. This term appears in the input and hidden layers of a neural network. Each node $z_j$, for $j \in\{0,1,2,3, \ldots, C\}$. In the hidden layer can be represented as a weighted sum of the nodes in the input layer:

where, C denotes the number of neuros in the hidden layer; $h^H$ is the activation function; and $w_{i j}$ represent the weights of the hidden layer. Finally, the node in the output layer, which we denote by y can be written as follows:

where, $h^0$ denotes the activation function for the output layer and $w_j^{\prime}$ represent the weights in the output layer. Hence, combining the above two equations we get:

There are a number of activation functions which include amongst others the logistic sigmoid, hyperbolic tangent and the softmax function. More details can be found in (H​a​y​k​i​n​,​ ​1​9​9​4). The values of the weights can be found using an algorithm called gradient descent method, in combination with the back and forward propagation. Several researchers have demonstrated that artificial neural networks are excellent at developing overall models, such as P​a​o​ ​(​2​0​0​8​) in modeling capital structure. However, neural network has not yet been applied to ascertaining the important explanatory variables affecting the FRL. Its use is particularly appropriate in this context, given the complexity of the data, which renders linear models unsuitable for capturing such intricate relationships.

Table 5 provides the sample distribution for our VO-year observations for all the 103 charities over the three years with a total of 309 observations. The table indicates that the highest proportion of filings, at 12.62%, occurred within the 241–270-day range. This suggests that a significant number of VOs submitted their accounts approximately 8–9 months after the end of the financial year, with the rate of submissions declining thereafter.

Table 6 presents the percentage of timely submissions for each year and category. The mean FRL in relation to Category 2 VOs, exceeds the permissible FRL during each of the years 2018, 2019 and 2020. The percentage of timely submissions in relation to this category ranges between 9.52% to 41.27%, with the year 2020 seeing an increase in timely reporting party due to the increase in the permissible FRL from 106 days to 180 days for that year, as noted in Table 2. In contrast, Category 3 VOs performed relatively well, with the mean FRL exceeding the permissible limit only in 2019. Relatively better results are also evident in the percentage of timely submissions which is consistently higher when compared to that of Category 2 VOs.

For both Category 2 and Category 3 VOs, the lowest rate of timely submissions was recorded in 2019, which may be attributed to the impact of the COVID-19 pandemic on their operations. An overall improvement in timely submissions occurred in the year 2020. This year, the highest percentage of timely submissions was recorded, with Category 2 also benefiting from an increase in the legally permissible FRL. This is in support of the initiatives introduced by OCVO, which aimed at strengthening its internal operations, thereby enabling the Commissioner to more effectively exercise regulatory oversight and carry out the vetting of VOs (O​C​V​O​,​ ​2​0​1​9). The legislative changes enacted during this time also arose from the need for VOs to uphold both upward accountability to funders and regulators, as well as downward accountability to beneficiaries, staff, and volunteers, both of which were examined in the literature, including that of D​h​a​n​a​n​i​ ​&​ ​C​o​n​n​o​l​l​y​ ​(​2​0​1​2​), O​’​D​w​y​e​r​ ​&​ ​U​n​e​r​m​a​n​ ​(​2​0​0​7​), and U​n​e​r​m​a​n​ ​&​ ​O​’​D​w​y​e​r​ ​(​2​0​0​6​). The improvement in the percentage of timely submissions also underscores the importance of good reporting in establishing trust (C​o​n​n​o​l​l​y​ ​&​ ​H​y​n​d​m​a​n​,​ ​2​0​1​3​a).

When compared to other countries, the situation in Malta appears less favourable. In the Belgian voluntary sector, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) report that in 2007 and 2008, approximately 17.2% of VOs filed their annual financial statements late, while C​h​r​i​s​t​i​a​e​n​s​ ​e​t​ ​a​l​.​ ​(​2​0​1​5​) observed that 24% of large and very large Belgian VOs failed to meet the statutory deadline during 2011 and 2012. Similar patterns are evident in England and Wales, where M​o​r​g​a​n​ ​(​2​0​1​1​) identified that 18% of registered charities submitted their accounts after the legal deadline. More recently, M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​), using data from UK charities between 2007 and 2018, found a considerably lower late filing rate of only 5.02%, suggesting a high level of compliance. By contrast, the timely submission rates of both Category 2 and Category 3 VOs in Malta during the period 2018–2020 were substantially lower, highlighting a persistent compliance challenge in comparison to the voluntary sectors of Belgium and the UK. Current trends further demonstrate progress in the UK, with the C​h​a​r​i​t​y​ ​C​o​m​m​i​s​s​i​o​n​ ​f​o​r​ ​E​n​g​l​a​n​d​ ​&​ ​W​a​l​e​s​ ​(​2​0​2​5​) reporting that 92% of charities filed on time by 2024–25, up from 81% in the previous year. A similar trend is, however, observed in Malta, where both Category 2 and Category 3 VOs demonstrated higher compliance in 2020 relative to the previous year. This aligns with the enactment of Legal Notice 317 of 2020, which amended the deadline for filing annual accounts and related documents with the OVCO, especially favouring Category 2 VOs. Nonetheless, the extent of timely submissions is still lower than that in other countries. This may be attributed to contextual and cultural factors, as well as the possibility that the reporting framework is overly burdensome for these VOs, as discussed in Section 2.1. Such findings also lend support to the reform proposals launched by OCVO in 2024 (M​i​n​i​s​t​r​y​ ​f​o​r​ ​I​n​c​l​u​s​i​o​n​ ​&​ ​t​h​e​ ​V​o​l​u​n​t​a​r​y​ ​S​e​c​t​o​r​,​ ​2​0​2​4), planning to almost double the thresholds and remove the requirement for Category 3 VOs to abide with IFRS, requiring them instead to abide with GAPSME.

Table 7 reports the descriptive statistics for the sample. The mean of FRL is 267.36 days which means that on average, VOs report between 8 to 9 months after their year end. A quarter of observations filed within 164 days and 50% of the VOs submit between 164 and 324 days. We also note that for SizeRevInc representing the size of the VO in terms of total revenue and income, the average revenue and income is €627,781. The median is €185,649, suggesting that most VOs are smaller, with some large VOs. The variable SizeAssets represents the size of a VO in terms of its total assets. The mean is €1,159,432 and the median is lower at €368,673, indicating some large VOs. Visual inspection of the data shows that 20% of the data relate to quite larger VOs and therefore form a distinct cluster, suggesting that they are not statistical outliers but a sub-group. RatioDG represents the extent of a VO’s reliance on donations and/or grants. A value closer to 1 suggests higher reliance on donations and/or grants, while a value closer to 0 indicates a more commercial VO. We note that the mean of 0.68 suggests that on average, the VOs lean towards reliance on donations and/or grants. This variable is useful for understanding the funding model of the VOs with a higher ratio indicating greater reliance on donations and/or grants.

The FRL affects the extent to which financial information is useful and it is therefore important to understand the main underlying factors affecting the FRL. We used non-parametric correlation tests because the data did not fully satisfy the assumptions required for parametric tests, in particular, the relationships between variables were non-linear.

(i) H1: Reliance on donations and/or grants

With respect to H1, Table 8 shows that the Spearman Correlation test yields a correlation coefficient of -0.107 and a P-value of 0.078. Although, as predicted, the negative coefficient indicates a tendency for VOs that primarily depend on donations and/or grants to exhibit a lower FRL, this relationship is not statistically significant. Therefore, H1 is not supported, suggesting that within the Maltese voluntary sector, reliance on donations and/or grants does not constitute a primary determinant of the FRL. This result contrasts with the correlation analysis of M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) and R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​), who both found the relationship to be statistically significant.

(ii) H2a: Size as total revenue and income

Moving on to H2a, as depicted in Table 9, when the size of the VO was measured by total revenue and income, a significant positive relationship was observed between the dependent variable FRL and total revenue and income at the 99% confidence level, supporting H2a. This aligns with the correlation analysis of prior studies, such as R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​), which suggest that larger entities may experience reporting delays due to increased complexity. On the other hand, the correlation analysis of M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) show a significant negative relationship. Empirical results therefore yield mixed results.

(iii) H2b: Size as total assets

As illustrated in Table 10, when total assets were employed as the proxy for size, no statistically significant relationship was found between the FRL and total assets, and therefore H2b is not supported. This result is consistent with the findings in C​h​r​i​s​t​i​a​e​n​s​ ​e​t​ ​a​l​.​ ​(​2​0​1​5​), who similarly found no association between size, measured by assets, and the FRL.

(iv) H3a: Profitability

The reporting of financial deficits in the voluntary sector may be perceived negatively (R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4; H​y​n​d​m​a​n​ ​&​ ​M​c​C​o​n​v​i​l​l​e​,​ ​2​0​1​8). In Belgium, R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​), through correlations, reported that that less profitable organisations may delay reporting. Looking at H3a, analysis of the voluntary sector in Malta using the Kruskal-Wallis test, depicted in Table 11, indicates a statistically significant difference in FRL across profitability groups; however, pairwise comparisons revealed no significant difference between profit- and loss-making VOs (p = 0.337), with significant differences observed only between break-even VOs and those reporting either a profit (p = 0.017) or a loss (p = 0.036). Since loss-making VOs did not significantly differ from profit-making VOs, the specific prediction about negative profitability is not fully supported.

(v) H3b, H3c, H3d: Liquidity, gearing, and equity

Turning to the remaining proxies related to bad news, both R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) and M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) report, based on their correlation analyses, that the filing of accounts is delayed in the presence of bad news represented by these three proxies, with these delays reported as statistically significant.

In the Maltese context, the Mann-Whitney U test between the FRL and liquidity (H3b) as illustrated in Table 12, indicates a statistically significant difference. VOs with a current ratio below 1 recorded a higher average FRL (300 days) compared to those above 1 (258 days), thereby supporting H3b and aligning with international evidence. By contrast, no significant associations were found between the FRL and either equity (H3c) or gearing (H3d), as illustrated in Table 13 and Table 14, meaning that these hypotheses are not supported. A plausible explanation is that Maltese VOs typically operate with limited access to external financing, rendering gearing less meaningful in this context. In such circumstances, liquidity constraints emerge as a more immediate and influential factor in shaping reporting behaviour, whereas equity and leverage appear to exert little practical influence on compliance with filing requirements.

(vi) H4: Year

Prior studies suggest that compliance improves over time as organisations adapt to regulatory requirements (O​w​u​s​u​-​A​n​s​a​h​,​ ​2​0​0​0), with L​u​y​p​a​e​r​t​ ​e​t​ ​a​l​.​ ​(​2​0​1​6​) finding evidence of a learning effect among Belgian firms. In this study, as illustrated in Table 15, a Kruskal-Wallis test revealed a statistically significant difference in FRL across the years, suggesting that the distribution of the FRL significantly differed across the years 2018–2020 (p < 0.001), consistent with the expectation of a decreasing trend over time, as also seen in Figure 2.

However, pairwise comparisons in Table 16 indicate no significant difference in the median FRL between 2018 and 2020 (p = 0.297). In contrast, the median FRL for 2019 differs significantly from both 2018 and 2020 (p < 0.001), reflecting substantially longer delays, as also indicated earlier, in Table 6. This can be attributed to the fact that the accounts for the year 2019 were processed in 2020, a period during which the COVID-19 pandemic disrupted administrative workflows and led to widespread filing delays. Although these results initially suggest an absence of a learning effect between the years 2018 and 2020, and that H4 is not supported, it is important to note that in the year 2020, the permissible FRL for Category 2 VOs was increased from 106 days to 180 days. This regulatory change likely influenced the observed patterns, as VOs were able to enjoy longer FRL periods. As a result, although the median FRL shows only a slight decrease between 2018 and 2020, this apparent stability may, in fact, not reflect a lack of a learning curve; instead, it may indicate that the extended permissible FRL allowed for longer processing times without violating policy limits, during a pandemic. Taken together, these findings highlight the need to consider both external disruptions, i.e., the COVID-19 pandemic, and regulatory adjustments when interpreting temporal trends in FRL, as they jointly shape the observed outcomes and may obscure underlying operational changes.

While the non-parametric tests indicate a few significant associations between predictors and the FRL, the neural network analysis captures more complex, non-linear interactions. We performed a neural network analysis to explore the combined, non-linear influence of the independent variables on the FRL. The dataset was divided into training (73.5%) and testing (26.5%) subsamples as illustrated in Table 17. The excluded 37 data points refer to missing data. The excluded 37 VOs represent those VOs having missing data. In the training part of the data, the model adjusts its internal weights and biases to minimize error.

The dependent variable was the natural logarithm of filing delay in days (LnDays), with predictors including profitability, equity, current ratio (liquidity), gearing, reliance of donations and grants, VO category, and year as shown in Table 18.

The model architecture, as shown in Figure 3, consists of a single hidden layer with three neurons, using a hyperbolic tangent activation function. This specification allows the model to capture non-linear relationships between the variables and late filing behaviour, thereby predicting the FRL. Each independent variable is represented as a node in the input layer. The lines connecting the input nodes to the hidden layer nodes are each assigned a weight (Synaptic Weight) which represents the strength and direction of the connection. The thickness and colour of these lines provide a visual representation of these weights depicted in Figure 3.

(a) Synaptic Weight > 0 (Grey lines): A grey line indicates a positive synaptic weight. The thicker the grey line, the stronger the positive relationship between the two nodes it connects. A positive weight means that as the value of the input variable increases, the activation of the hidden node also tends to increase.

(b) Synaptic Weight < 0 (Blue lines): A blue line indicates a negative synaptic weight. A thicker blue line signifies a stronger negative relationship. A negative weight means that as the value of the input variable increases, the activation of the hidden node tends to decrease.

In Figure 3, the visual representation of the weights, particularly the thickness and colour of the lines, allows for a visual understanding of which predictors have the most influence within the neural network’s structure, reinforcing the findings presented in Table 19.

Table 20 shows that performance statistics indicate strong predictive power without overfitting: the relative error is 0.891 in training and 0.815 in testing, with both values being low and close.

The analysis of the model’s output in Table 19 reveals the relative importance of each independent variable in predicting the FRL. Table 19 indicates that the Year variable is the most influential (importance = 0.259; 100% normalised importance). This is followed by Profit (65.8%) and Equity (63.8%). RatioDG, representing the dependency on donations and grants (51.9%), and liquidity, measured by the current ratio (CR) (45.6%), exhibit a moderate influence. The VO Category (29.9%) and Gearing (29.2%) report the lowest importance scores, suggesting their limited influence on the neural network’s predictions.

We analysed the thickness and colour of the connections from each input variable through the hidden layer to the output node, in Figure 3, to infer the overall predicted sign of each variable’s relationship with the FRL. This was examined together with the independent variable importance reported in Table 19. The most influential variables are analysed next, together with the results of non-parametric tests presented in Section 4.2.2.

Year emerged as the most influential variable in the model. This is further illustrated in Figure 3, where the connections from the Year nodes to the hidden layer are the thickest, indicating that the variable is of high importance. Taking 2018 as the base year, the connections for 2019 are represented by thick grey lines, indicating that the related filings were associated with longer FRLs compared to those related to 2018. This increase aligns with the pairwise comparisons reported in Table 16, which indicate that median FRL in 2019 is significantly higher than in both 2018 and 2020, likely reflecting administrative delays caused by the onset of the COVID-19 pandemic.

In contrast, the year 2020 is represented by predominantly thinner blue lines in Figure 3, suggesting that filings related to this year had shorter FRLs relative to the year 2018. The non-parametric test results indicate no significant difference in median FRL between the years 2018 and 2020. The apparent stability in 2020, despite a potential learning curve, may be influenced by the regulatory increase in the permissible FRL for Category 2 VOs from 106 days to 180 days, allowing for more extended processing periods without breaching policy limits.

Taken together, these findings indicate that the neural network, which identifies Year as the most influential variable with 100% normalised importance, effectively captures temporal variations in FRL, including the 2019 pandemic-related delays and the combined effects of policy changes in the year 2020 and potential learning effects during this period.

Profitability is the second most influential variable in the neural network model. While the non-parametric test found no significant difference in FRL between profit- and loss-making VOs, the neural network analysis strongly suggests a negative relationship for loss-making VOs, as indicated by the thick blue connections to the hidden layer. This finding suggests that VOs reporting a loss may, contrary to the literature (M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​,​ ​2​0​2​4; R​e​h​e​u​l​ ​e​t​ ​a​l​.​,​ ​2​0​1​4), file their accounts more promptly than break-even or profit-making VOs. This seemingly counterintuitive result could be interpreted through the lens of signalling theory and organisational legitimacy. When faced with a financial deficit, a VO’s reputation and continued access to funding sources become highly vulnerable. In such a scenario, prompt filing, despite the unfavourable results and the seemingly bad news, could be a strategic move to signal transparency and accountability to stakeholders. By doing so, the VO demonstrates that it is not concealing bad news. This approach can help mitigate reputational damage and maintain the trust required to secure future funding. Therefore, rather than a cause for delay, a financial loss may act as an incentive for timely reporting.

The neural network analysis in Table 19 reports CR (liquidity), Equity, and Gearing as potential predictors of the FRL, with Gearing showing the least importance. Consistent with the Mann-Whitney U test, Gearing is not a significant predictor of the FRL, and CR is reported as a significant factor, albeit with lower normalised importance compared to other predictors. This convergence reinforces the evidence that liquidity constraints are associated with longer reporting delays. The latter is also consistent with the regression analyses in R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) and M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​), where liquidity is statistically significant in both studies.

Although the Mann-Whitney U test found no statistically significant association between Equity and the FRL, the neural network assigned Equity a relatively high normalised importance. This suggests that the neural network is uncovering a subtle yet influential effect that is masked in a univariate context. This mixed evidence mirrors prior international findings. R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) report that equity is not statistically significant in regression models, whereas M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) identify equity as a significant determinant of reporting timeliness. Taken together, these results suggest that the role of equity may be context-dependent.

With respect to RatioDG (Reliance on donations and/or grants), while the Spearman correlation indicated a negative association with FRL, which is not statistically significant, the neural network assigns RatioDG a normalised importance of 51.9%, suggesting that this variable plays a moderate role in predicting FRL when considered in conjunction with other determinants. This result aligns with the regression analysis of M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) and R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​), who both found the relationship to be statistically significant, offering support for the resource dependence theory, which posits that donative VOs seek to adhere to legal and regulatory requirements, including the timely submission of reports (V​e​r​b​r​u​g​g​e​n​ ​e​t​ ​a​l​.​,​ ​2​0​1​1). The results also support the signalling theory, which suggests that donors are more likely to continue funding non-profit organisations that demonstrate accountability, reflected, for example, in the timely filing of accounts (F​e​l​i​x​ ​e​t​ ​a​l​.​,​ ​2​0​1​7; H​y​n​d​m​a​n​ ​&​ ​J​o​n​e​s​,​ ​2​0​1​1; H​y​n​d​m​a​n​ ​&​ ​M​c​K​i​l​l​o​p​,​ ​2​0​1​8).

With respect to SizeRevInc (size measured by total revenue and income) and SizeAssets (size measured by total assets), similar to prior empirical results, the non-parametric tests yielded mixed results. Regression results reported in R​e​h​e​u​l​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) show no statistically significant relationship. In contrast, M​a​y​a​p​a​d​a​ ​e​t​ ​a​l​.​ ​(​2​0​2​4​) found a significant positive relationship when size is measured in terms of income and a significant negative relationship when size is measured in terms of assets. Interestingly, the neural network model did not identify either size variable as a key predictor of FRL. These mixed empirical findings reinforce argument of B​e​a​t​t​i​e​ ​e​t​ ​a​l​.​ ​(​2​0​0​1​) argument that measuring size in the voluntary sector is inherently complex and subject to variation depending also on the proxy applied. In the neural network, size is additionally represented by the variable Category, distinguishing between Category 2 and Category 3 VOs. Consistent with the mixed results observed in the non-parametric tests, this variable exhibits a low normalised importance of 29.9%.

This is the first comprehensive empirical study of reporting timeliness in the voluntary sector in Malta. Our analysis is based on a sample of 103 VOs from 2018 to 2020. It also identifies the important explanatory variables that affect the VOs’ financial reporting timeliness.

The application of a neural network in this study adds methodological value to research on the FRL in the voluntary sector. Unlike traditional regression models, neural networks capture complex, non-linear interactions between predictors, which allows for a better understanding of how variables jointly influence compliance. This approach adds value and extends the methodological toolkit available to researchers, yielding insights that linear models may overlook.

The results confirm Year as the most influential predictor, with the neural network capturing pandemic-related delays in 2019 and the interplay of policy changes and learning effects in 2020. Profitability and equity emerge as the strongest explanatory variables. Moderate contributions are observed for dependency on donations and grants, as well as for liquidity. In contrast, the VO category and gearing exhibit the lowest importance scores, indicating a limited role in shaping the neural network’s predictions.

The analysis further indicates that Category 2 VOs persistently exceeded the statutory reporting lag from 2018 to 2020, with compliance improving only in 2020 following the extension of filing deadlines. In contrast, Category 3 VOs generally demonstrated better timeliness, although compliance deteriorated in 2019, mainly attributable to COVID-19 disruptions. Compared to Belgium and the UK, compliance levels in Malta remain substantially lower, highlighting ongoing challenges.

The findings carry several implications for policy and practice. The OCVO should ensure consistency in monitoring and enforcement across years to minimise fluctuations in the timely filing of accounts. Particularly in exceptional circumstances, such as the COVID-19 pandemic, the OCVO could also offer support that helps mitigate resource-related barriers. Additionally, simplified filing systems, training, or administrative assistance may improve compliance outcomes. This is especially important when considering that some VOs are subject to relatively more complex financial reporting requirements compared to their for-profit counterparts, which use GAPSME, and which could ultimately result in financial reporting filing delays.

This study is particularly timely in view of the new set of proposals launched by the OCVO in 2024 to reform the voluntary sector and reduce the relative obligations it is subject to. Addressing the call for further research on this sector by the Ministry for Inclusion & the Voluntary Sector, this study provides evidence to inform decision-making by the government, the OCVO and other stakeholders.