Requirements for the annual financial statements in the Netherlands are enclosed in Title 9 Book 2 of the Civil Code. According to article 2:384 of the Civil Code financial statements are prepared based on the going concern assumption, unless there is reason to assume other­wise. The Civil Code states; ‘The valuation of assets and liabilities shall be based on the assumption that all the activities of the legal person to which those assets and liabilities are helpful, are to be continued unless that assumption is incorrect or its accuracy is subject to reasonable doubt; then this will be clarified, with mention of the impact thereof on capital and result’. Therefore, an explanation in the financial statements is obligatory when the assumption of continuity is no longer valid or there is reasonable doubt about its correctness. Figure 1 depicts a scheme that shows the requirements under the Dutch Civil Code for going concern reporting.

Figure 1.

Accounting regulations for going concern under Dutch Civil Code.

The legislator gave further explanation of the legal provision stated above, in an Explanatory Memorandum. The explanation states (Parliamentary Papers II, 1979–1980, 16326 nr. 3, p. 21.): ‘When continuity is not assured, this should be expressed in an explanatory statement; in that case, reason exists to valuate assets and liabilities on another base than usual, such as liquidation base. This course of action follows for the guidance of article 2 paragraph 5; in accordance with this provision it is required to describe the effect on equity and results.’. Therefore, an explanatory statement in the financial statements is obliged when continuity is not assured. In addition, one can conclude from the Explanatory Memorandum that in a situation when continuity is not assured, there is no exact prescription for the wording and content of that explanatory statement. This is different in the situation when going concern is not assured and another valuation base is applied. In this situation, the legislator prescribes to present the effect on equity and results. In short, both law and explanatory memorandum do not state explicitly what is meant by ‘reasonable doubt’.

The Dutch Accounting Standards Board – Raad voor de Jaarverslaggeving (RJ) – develops and publishes standards that explain the Civil Code. Standard A2.214 describes a situation of ‘reasonable doubt’ as follows (free translation); ‘Reasonable about the continuity of the activities of the legal entity exists when the legal entity is no longer able to fulfill its obligations under own control. This means that the going concern of the legal entity is inevitable without further support of stakeholders to a larger extent than they are obliged to, while it is not sure if this further support would be provided. The standard furthermore indicates that ‘reasonable doubt’ anyhow exists in a situation where a company is not able to fulfill obligations under its own control. To answer the question of whether a business will no be able or not to fulfill its obligations under its own control it is particularly important to review the liquidity position of the legal entity. In line with the aforementioned, both the professional member organization of auditors (NBA) as the Dutch Accounting Standards board (RJ) in the Netherlands have frequently drawn attention to the liquidity position of companies in times of great uncertainty due to the COVID-19 pandemic crisis (NBA 2020; RJ 2020).

In this study, we use two different databases. The selection criteria are summarized in Table 1. First, we identified all limited liability companies that are subject to a statutory audit in the Netherlands in the database via Company.info. This is a commercial company that discloses financial information of companies in the Netherlands. Financial statement information in the database is collected continuously via the Chamber of Commerce. After selection based on the aforementioned criteria, we excluded:

• public companies;
• companies that report on another standard than Title 9 Book 2 Civil Code (i.e. IFRS) because we are primarily interested in the application of the Dutch Civil Code;
• all companies with a fiscal year-end date after COVID-19 outbreak as were included in their financial statements (for instance fiscal year-end March 31, 2020);
• companies that filed the financial statements before the COVID-19 outbreak, since the effects of COVID-19 could not be included in these financial statements (for instance a company with a fiscal year-end date December 31, 2019 publishing before COVID-19 outbreak).

Next, we randomly drew a sample of 760 companies. The sample selection criteria are summarized in Table 2. Based on the selection criteria for this study the minimum sample size is 373. However, we increased the initial sample because some variables included in the model, see paragraph 3.2, are not included in the financial statements. For example, according to the Dutch Accounting Standards, a cashflow statement is mandatory for medium and large-sized companies. However, prior research shows that it is not uncommon that companies apply for an exemption to include a cashflow statement or omit the cashflow statement without disclosure for a valid reason (Vergoossen and Meershoek 2018). Another example is that the preparation date of the financial statement is not given in every financial statement (Vergoossen and Van Beest 2019).

From all the companies in the sample, we collected the annual financial statements and we manually parsed the data from these financial statements. For the application for a government grant, we use the data register of the UWV (the Employee Insurance Agency in the Netherlands). This institute publishes data on the application for government grants. In total 181 companies were removed from the sample when parsing data failed or data were not included in the financial statements. This results in a total of 579 companies being included in the dataset for this study.

Following prior research (e.g. Fargher and Jiang 2008; Berglund et al. 2018; Hardies et al. 2018), we use the following logistic regression model to explain the probability of a mandatory going concern disclosure:

GC-FS = β0 + β1 GRANT + β2 NCF + β3 CFO + β4 CASH + β5 SIZE + β6 LEV + β7 ROA + β8 BIGN + ε

The variables used in this research are defined in Table 3. The dependent variable (GC-FS) is a dummy variable equal to 1 (one) if the financial statements contain a mandatory explanatory statement on going concern uncertainty, and equal to 0 (zero) otherwise.

We use two types of test variables, being government grants and liquidity indicators. We use three different variables for a government grant (GRANT). GRANT is a dummy variable equaling to 1 (one) if a company applied for a government grant before the preparation date of the financial statements, and 0 (zero) otherwise. N-GRANT is the total number of applications for a government grant before the preparation date of the financial statements. A-GRANT is a dummy variable equaling to 1 (one) if a company applied for all government grants before the preparation date of the financial statements, and 0 (zero) otherwise. We use three proxies for liquidity indicators in this study. NCF is a dummy variable equal to 1 (one) if the cashflow is negative, and 0 (zero) otherwise. CFO is calculated as the ratio between cashflow from operations and total liabilities. CASH is calculated as the total sum of cash and cash equivalents divided by total assets.

We include multiple variables to control for factors that have previously been shown to be associated with going concern uncertainty disclosures in the financial statements. SIZE is calculated as the natural log of total assets. LEV is calculated as the ratio between total liabilities and total assets. ROA is calculated as the ratio between the net result and total assets. BIGN is a dummy variable equaling to 1 (one) if the auditor is a Big 4 audit firm, and 0 (zero) otherwise.

In Table 4 we present both the descriptive for the full sample and the univariate statistics for the independent groups. Table 4 presents the descriptive statistics for the full sample of companies (N = 579). In the sample of this study, 7% of the financial statements contain mandatory going concern expressions (GC-FS). This percentage is in line with prior research (Knechel and Vanstraelen 2007). As for the test variables, in total 20% of the companies applied for a government grant (GRANT) before preparation of the financial statements. As would be expected, fewer companies applied for all three grants (A-GRANT). In total 2% of the companies applied for all three grants. Furthermore, on average companies applied for 0.3 numbers of grants (N-GRANT). This value shows that on average companies that did apply for a grant on average opt 1.3 times. 41% of the companies present negative cashflows (NCF), on average 55% of the liabilities can be covered by operating cashflows (CFO) and on average 12% of the total assets contain cash of cash equivalents (CASH). The mean values of the control variables are comparable with prior research; the log of the total assets is 17.27 (SIZE), 57% of the balance sheet contain liabilities (LEV), the return on assets is 6% (ROA) and 45% of the companies in this sample is audited by a Big 4 firm.

In addition, we performed the non-parametric Mann-Whitney test to compare the means of the two independent groups. Table 4 highlights some basic differences between the companies that expressed a disclosure and companies that did not express a disclosure on going concern in the financial statements. It appears that all the variables included in the model are significant factors determining the likelihood of a disclosure. Only the size of the company (SIZE) and when companies apply for all the grants in the year (A-GRANT) are not significant.

We identified items that potentially could indicate an outlier by verifying the standard deviation from mean values. We assessed each item and did not exclude any identified items because they all match with the selection criteria for this study. All the values of the aforementioned independent variables, test variables, and control variables are comparable with prior research (e.g. Knechel and Vanstraelen 2007; Fargher and Jiang 2008; Hossein et al. 2018).

The Pearson correlation matrix is shown in Table 5. The correlations that are shown concern both continuous variables and dichotomous variables. The correlations between the variables that reflect the application for a government grant (GRANT; A-GRANT and N-GRANT) vary between low and high. We do not combine these variables in a regression model. Other variables included in the research model correlate moderately at most. The correlation between the type of auditor (BIGN) and the size of the client (SIZE) is 0.455 and the correlation between the return on assets (ROA) and the ratio between operating cashflow and total liabilities (CFO) is 0.594. All other correlations between the independent variables are substantially lower. The aforementioned correlation levels are comparable with other studies on going concern uncertainty (i.e. Hardies et al. 2018). This indicates that only a very limited coherence exists between the independent variables. In addition to the Pearson correlation matrix, we performed VIF analyses and find low values. Based on the aforementioned we conclude there is no problem with multicollinearity.

In Table 6 the results of the logistic regression analysis are shown. The dependent variable is the going concern uncertainty disclosure in the financial statements (GC-FS). In the basic model (Model 1) we only include the control variables which are used in prior research. In the models thereafter (Model 2, 3, 4 and 5) we added the test variables for this study and conclude with the full model (Model 6) in which all variables are included.

Hypothesis 1a focuses on the association between negative cashflow (NCF) and the explanatory statement on going concern uncertainty in financial statements (GC-FS). Results in Table 6 show that a going concern uncertainty disclosure is more likely for companies with negative cashflows (NCF) that show positive and significant results (Model 3: coefficient 0.730, p < .05, model 6: coefficient 0.625, p. < .05). Therefore the result is consistent with H1a. Hypothesis 1b focuses on the association between the operational cashflow and total liability ratio (CFO) and the explanatory statement on going concern uncertainty in financial statements (GC-FS). Results in Table 6 show that a going concern uncertainty disclosure is less likely for companies in which operational cashflow relative to liabilities is larger. CFO shows negative and significant results (Model 4: coefficient -0.914, p < .05, Model 6: coefficient -0.839, p < .05). Therefore the result is consistent with H1b. Finally, Hypotheses 1c focuses on the association between the ratio between the size of liquid assets and total assets (CASH) and the explanatory statement on going concern uncertainty in financial statements (GC-FS). Results in Table 6 show that a going concern uncertainty disclosure is less likely for companies with higher liquid assets re­lative to their total assets. CASH shows negative and (in)significant results (Model 5: coefficient -1.876, p < .05, Model 6: coefficient -1.296 p > .10). Therefore the result is not consistent with H1c.

Hypotheses 2 focuses on the association between the application for a government grant (GRANT) and the explanatory statement on going concern uncertainty in financial statements (GC-FS). Results in Table 6 show that a going concern uncertainty disclosure is more likely for companies that applied for a government grant. GRANT shows positive and significant results (Model 2: coefficient 0.887, p < .01, Model 6: coefficient 0.900, p. < 0.01). The effect of an application of a government grant (GRANT) is positive and significant in all the regression models shown in Table 6. This indicates that the application for a government grant increases the propensity of issuing a going concern uncertainty disclosure in the financial statement. Therefore the result is consistent with H2.

We performed several additional analyses on the association between the government grants and going concern uncertainty. In the basic model (see Table 6) an application for a government grant (GRANT) is defined as a company applying for a single grant before management prepared the financial statements. In the additional analyses, we substituted the primary test variable (GRANT) for the total number of applications for a government grant in2020 before the preparation date of the financial statements (N-GRANT). Results in Table 7 show that a going concern uncertainty disclosure is more likely for companies with more grant applications. N-GRANT shows positive and significant results (coefficient 0.454, p < .05). Next, a company can apply for a maximum of three grants (A-GRANT). A-GRANT will be assigned value 1 (one) if the company before the preparation date of the financial statements applied for all three available grant programs in the year 2020, and 0 (zero) otherwise. A going concern uncertainty disclosure is more likely for companies that applied for all three available grant programs in the year 2020. A-GRANT shows positive, but no significant results (coefficient 0.901, p > .10).

In addition to the basic model, we replaced the dependent variable and then investigate the association between multiple liquidity indicators and the application for a government grant (GRANT), and the application for the number of grants (N-GRANT). We expect that lower levels of liquidity, directly drawn from the financial statements, also affect the application for a government grant (GRANT), and the application for the number of grants (N-GRANT) in the period thereafter. In Table 8 the results for the logistic and linear regressions are shown. We do not find any significant results for negative cashflows (NCF), the ratio between operational cashflows and liabilities (CFO), and the total of cash to total assets ratio (CASH). These results indicate that liquidity indicators prior to COVID-19 do not affect the application for the government grant.