Dividend policy and crisis: Exploring the interplay between performance and financial constraints in the French context

Abstract

The purpose of this paper is to determine the effect of the Covid-19 pandemic crisis on dividend policy and performance and explore the interplay between performance and financial constraints to identify how such a fit affected dividend policy during the crisis.

We used a final sample of 106 SBF-listed firms during six years. To assess the effect of the crisis, we divided this period into three subperiods: pre-crisis (2016–2018), pandemic period (2019–2021), and all periods (2016–2021). A System Generalized Method of Moments (SGMM) is used to deal with the problem of endogeneity caused by the lagged dependent variable.

The results showed that only the crisis and the financial constraints (KZ index) negatively correlated with dividend payment levels (DivPaid). This dividend level did not take performance into account. Regarding the control variables, only debt, growth, and size positively impacted dividend levels. Moreover, the performance of French companies was negatively influenced by the DividPaid, KZindex, and Crisis variables. The findings suggest that France should prioritize dividend payments to protect a company's reputation and financial health. These findings have significant implications for investors, financial analysts, regulators, and policymakers who are looking for guidance on dividend policy in uncertain situations.

1. Introduction

The covid19 pandemic had repercussions on almost all areas of activity. As such, it induced policymakers to find a solution and design a policy to cope with the aftermath of this crisis. Notably, corporate managers in different countries (UK, Germany, France, Italy, Canada, …) decided to reduce dividend distribution rates or even cancel them to convey positive information about firms’ long-run growth prospects [1]. During the covid 19 crisis, companies suffered from lower profits, higher earnings volatility, higher return volatility, and deteriorating stock prices [2].

Indeed, the economic downturn, lack of visibility, and government pressure are forcing firms to preserve liquidity to guarantee a better solidity of their balance sheets. Several studies showed that the impact of covid 19 was more pronounced on the returns of firms that did not have liquidity reserves than those that had more cash reserves [3]. Firms with more cash reserves had better operating performance and less investment [4,5]. Accordingly, cash holdings helped mitigate the negative impact of covid 19 shocks on corporate payouts [1]. [6] showed that financial and/or economic crises have consequences for firms' financial policies in subsequent years. Crises represent exogenous shocks to the performance, sustainability, and cash flows of firms. Periods of uncertainty reduce the expected profitability of investments, which increases the sensitivity to cash flows and the level of financial constraints of firms [7]. This finding is also observed by Refs. [2,8]. These authors concluded that firms were financially constrained by the covid 19 pandemic. Financial constraints are considered as obstacles to optimal investments, dividend payout and destructive to firm performance [9]. They intensify when a firm suffers from a lack of internal funds and is unable to attract external funds. As a result, the firm is unable to finance value-creating investment opportunities and has to choose between worthwhile investments [10].

However, some studies separately examined dividend policy and firm performance during a crisis period. The results pointed to the negative impact of a crisis on performance [[11], [12], [13]] and on dividends [2]; Leirvik, 2022; [1,14]. To the best of our knowledge, few studies have examined the direct relationship between dividend policy and firm performance during a crisis period [15,16]. have examined this relation in the Covid-19 and subprime crisis periods. Our study discusses the interplay between dividend policy and performance taking into account the financial constraints generated during the covid 19 crisis in a sample of French firms listed on the SBF120 during the period 2016–2021. We divided the period into 3-year sub-periods; all periods (2016–2021), the pre-pandemic period (2016–2018) and the pandemic period (2019–2021) to show the effect of this crisis. The System Generalized Method of Moments (SGMM) is used to study the complementarity or the subsidiarity between dividend payment and financial performance.

We chose France as the context of this research for several reasons: The first reason is that previous research on the relationship between financial constraints and dividend policy were conducted almost exclusively in the United States and Chinese contexts (Rent et al., 2021). Also, the French context has several peculiarities, in terms of the legal regime and governance mechanisms. Indeed, several studies showed that the ownership structure of French companies differed from that of American and Canadian companies [16]. According to Ref. [17]; the ownership structure of publicly traded firms in France is more concentrated than in the United States where it is highly dispersed and has different governance characteristics. This difference in ownership structure may lead to different behaviors of managers in terms of financial constraints [18].

The second reason is that this period is known by the launching of corporate governance and anti-corruption policies following the new laws and international tax standards that were implemented in both the United States and France (Rent et al., 2021). These latter policies had an impact on the behavior of managers in terms of financial policy.

The results of the study have significant implications. First, they provide a better understanding of how companies manage their dividend policy during times of crisis and shed light on the consequences of this policy on their performance and financial constraints. Second, the study offers valuable information for policymakers and regulators, enabling them to encourage companies to adopt a stable dividend policy during crisis periods. Lastly, the findings of the study can also help investors and financial analysts to better understand the factors that influence the dividend decisions of French companies and to assess the performance and financial stability of these companies.

The remainder of this paper is structured as follows: Section 2 reviews the relevant literature and develops the research hypotheses. Section 3 describes the research methodology. Section 4 reports and discusses the results. Section 5 concludes the paper with the research implications, limitations, and future research.

2. Literature review and hypotheses development

Many theories explain the relevance or irrelevance of dividend policy in determining the value of firms [19]. First, the theory of dividend irrelevance, proposed by Miller and Modigliani (1961) suggests that shareholders are generally indifferent between dividends and capital gains. This theory is also advanced by Ref. [20] and shows that dividends may contribute to the sustainability of firms in the capital market [21]. Also, the bird-in-hand theory suggests that companies paying dividends are seen as more attractive to investors. Second, the theory of free cash-flow argues that paying a portion of surplus cash flows, after funding all profitable and positive-yielding investment opportunities, helps mitigate agency conflicts and reduce agency costs (Hasan et al., 2021; [19,22]. Third, the pecking order theory of dividends, proposed by Ref. [23]; suggests that companies prioritize using retained earnings to fund investments. While the pecking order theory does not directly address dividends [24], argue that this theory is relevant for reconciling dividend decisions with investment needs.

Generally, dividend policy takes a grand position under the signaling and agency theory. It's considered an important pillar of corporate finance and one of the most important financial decisions that may affect a company's performance [5,25]. This impact is related to various factors such as the economic conditions of countries [2,[26], [27], [28]]and financial constraints faced by the company [29]; Almeida et al., 2004; Hribar & Collins, 2002; [30].

2.1. Dividend policy and financial performance

Several studies demonstrated a bi-directional relationship between performance and dividend policy.

[25] studied the impact of performance on dividend pay-out during 12 years in China. The performance is measured by Return On Asset (ROA) and Return On Equity (ROE) and the dividend policy by the Dividend Payout Ratio (DPR). The results showed a strong positive relationship between DPR and the two measures of performance. This finding supports the research of [31] whose study detected a strong link between dividend policy and corporate financial performance in emerging countries. The author studied 92 industrial and service sector companies listed on the Amman Stock Exchange observed during the period stretching from 2015 to 2019, using a Panel Data Analysis. The results indicate that dividend policy explains much of a company's financial performance.

High profitability leads to high cash flows and can therefore lead to generous dividend payments [32,33]. The company can decide to keep its profit for future reinvestment. Hence, dividend policy depends on the current or future earnings of the companies. Assessing this assumption, many researchers found a positive relationship between dividend payments and current earnings, especially in developing economies [34]. Indeed [8], studied the relationship between dividend policy and earnings in view of the role of agency problems and financing constraints. Their findings show that firms with higher profitability pay high amounts of dividends and consequently retain few earnings.

Several studies, such as those of [24,35]; showed that dividend payment positively relates to income earned by the company. A dividend payment can help improve business performance by reducing agency conflicts. Referring to signaling theory [36], suggested that revenues were positively associated with dividend payment in the Tunisian context. This study showed that listed firms distributed dividends to convey information to shareholders about the firm's health, and more generally to other stock market participants. Thus, dividends inform the market about the firm's present and/or future cash flows. They give investors a signal about the firm's earning power.

In the light of these proposals, we developed two hypotheses linking performance and dividend policy:

H1

Dividend payment has an impact on performance.

H2

Performance has an impact on dividend policy.

2.2. Dividend policy and financial constraints

Based on the hierarchical order hypothesis, the relationship between dividend policy and financial constraint is inverted. Miller and Modigliani (1961) and [23] argued that the company paid dividends after having satisfied financing all investment projects.

According to Almeida et al. (2004), financially constrained firms might experience a decrease in liquidity if they increased their dividend payouts. This decrease may then lead to a reduction in the firm's investment activity, which could ultimately affect its position in the market. Firms that are financially constrained are thought to face costs that are significantly high to attract external funds. Moreover, they are linked to a high information asymmetry between investors and firm management [7].

[37] argued that there is no single comprehensive theory of dividend payout. They indicated that even firms with difficult access to external financial resources might decide to distribute dividends and induce investors into predicting dividend payment as a positive signal of firm performance [32]. [38] empirically examined a sample of Italian SMEs between 2015 and 2019 to determine the relationship between dividend payout policy and financial constraints, using credit ratings as a measure of access to external financial resources. They found a positive relationship between firm solvency and dividend payout, suggesting that firms with high financial constraints are less likely to pay out dividends.

Moreover, taking into consideration the fundamentals of financial constraints theory proposed by Refs. [14,39] affirmed that a firm's investment should be proportional to the value gained by the potential investment opportunities and independent of the company's financial status when the financial market is unconstrained. Hence, the following hypotheses are formulated:

H3

Financial constraints have a negative impact on dividend policy.

H4

Financial constraints have a negative impact on financial performance.

2.3. Dividend policy during a crisis period

Dividend policy is not only determined by firm characteristics but also by the business environment, such as a crisis period. Previous research revealed a downward trend for dividend payouts in times of financial crises (Mili et al., 2017). These studies indicated a positive relationship between governance mechanisms and paid dividends in periods of financial crises. This is explained by a high ownership concentration and the dominance of institutional investors [15]. found evidence of the role played by ownership structure and board characteristics in explaining the behavior of dividend policy of Tunisian firms during a crisis. They proved that firms with a higher proportion of institutional shareholders have higher dividend payout ratios, and the combination of the positions of chairman of the board and chief executive officer has an effect on the profit distribution decision.

Building on the work of [24,26,40]; Boussman et al. (2022) disclosed that the percentage of firms distributing dividends declined between 2008 and 2009 (a statistically significant decline at 1 %) and that the distribution policy varied considerably during the crisis (study period 2006–2009). Moreover [41], showed that the financial crisis in the Istanbul market in 2001 led to a significant decline in the number of firms paying dividends. Such a decline in dividends during the 2008–2009 financial crisis is confirmed by Ref. [42] in European banks [2]. tested the effect of the COVID-19 pandemic on corporate dividend policy. Examining a sample of the G-12 countries, the findings show that although the proportion of dividend cuts and omissions is significantly higher during the pandemic, most firms maintain or decrease the distributed dividends percentage. The company pursues a stable dividend policy and signals its financial prospects during a crisis, as proposed by dividend signaling theory. Moreover, the results showed that firm profitability, earnings prospects, size, and leverage appear to be important determinants of dividend policy decisions during the pandemic [2].

Bearing on the above findings, and taking into account the latest covid 19 pandemic, we formulate the following hypothesis:

H5

The covid 19 crisis negatively impacted dividend policy and performance.

3. Research design

3.1. Sample and collection

To explore the relationship between performance, dividend policy, and financial constraints we examined a sample extracted from the Data-Stream database and the annual reports of the firms listed on the SBF 120 index. The firms have the most significant stock exchange capitalization. Of the companies making up the SBF120 index, we eliminated foreign companies as well as companies whose data was missing. This reduced our final sample to 106 French companies operating in different sectors, observed during the period 2016–2021.

To determine the effect of the crisis, we divided the period into two sub-periods of three years each: the entire study period (2016–2021); before the health crisis (2016–2018), and during the health crisis (2019–2021). This distinction allows us to compare the performance of companies according to the adopted dividend distribution policy.

Furthermore, we classified the sample into companies that continued to operate during the covid 19 lockdown period. First, we referred to ICB classification to specify more industry observations (industry classification/ten industries). Second, we opted for the classification used by Bozzolan et al. (2006) and divided our sample into two groups: one represents the traditional industry (includes sectors such as food, automobiles, chemicals, construction, electronics, manufacturing, oil, utilities, textile/clothing, and tourism/leisure) and the other the high-tech industry (includes sectors such as internet application provision, biotechnology, entertainment, IT, retailing, high tech manufacturing, media, software, systems integration, telecommunications, and web services). We also distinguished tech companies that continued to operate during the covid 19 lockdown period.

Table 1 describes the sample by sector. It shows the traditional and high-tech industries, representing respectively, 57.5% and 42.4%. of the total sample. Moreover, with the ICB classification, industrial activities are the most represented sector in the total sample, with 26.4 %.

Table 1.

Sample distribution by sectors.^a.

Industry	Firms' Number	Percentage of Firms	Industrial	Firms' Number	Percentage of Firms
Oil & Gas	5	4.7	Traditional industry	61	57.54
Basic Materials	5	4.7
Industrial	28	26.4
Consumer Goods	20	18.86
Health Care	5	4.7
Consumer services	17	13.2	High-tech industry	45	42.45
Telecommunications	5	4.7
Utilities	5	4.7
Technology	17	16.03
Total	106	100 %		106	100 %

^a[43] and ICB Classification.

Besides, we divided our sample into three categories according to the type of distribution decision taken during the crisis (Table 2): canceled, reduced, and continued distributions of dividends. We note from Table 2 that only 20.75 % of the total sample continued to pay dividends against 41.5 % that canceled this type of payment.

Table 2.

Distribution of the sample according to the decision types during the covid 19 crisis.

Decision types	Number	Percentage (%)
Canceled dividend distribution	44	41.5
Reduced dividend distribution	40	37.7
Retained dividend distribution	22	20.75
Total	106	100

Fig. 1 gives an idea of the nature of the industry with a high dividend payment (DividPaid). This figure shows that the Oil and Gas industry pays on average 40.05, followed by consumer services (34.011) and telecommunications (28.9).

Figure 1.

DivPaid Variable by sectors.

3.2. Variables selection

As in previous studies, we selected the following variables:

Financial constraints: it is measured in the literature by several indexes such as the KZ index of [44]. We calculated the KZ for each firm-year observation. According to the Agency Theory, firms with higher financial constraints tend to have higher agency costs due to the conflicts between managers and shareholders. A dividend policy reduces these conflicts.

Dividend paid: it is calculated as dividend payout per share (Nantamins and Zhou 2022). If a company pays dividends, it is considered not financially constrained and if it does not pay, it is classified as financially constrained.

Performance: it is measured by two variables: return on assets (ROA) and return on equity (ROE). ROA is defined as the net income to total assets and indicates whether a firm efficiently uses its assets to generate revenues. ROE is defined as the net income to total equity and assesses financial returns to shareholders [31,45].

Cash flow: it is measured by the cash flow to total assets ratio to estimate the quality of a company's earnings. We expect its positive impact on dividend policy (Brealey et al., 2008).

Firm size: it is the natural logarithm of total assets [40]. indicated that firm size is positively related to the dividend payout ratio. In addition, large companies should be less financially constrained than small companies.

Debt: it is defined as the ratio of total debt to total assets. According to Agency Theory, we expect a negative correlation between debt and the dividend payout ratio [15].

Age: it is measured by the number of years since the company has been founded [24].

Sales growth: it is measured by the rate of change in sales scaled by the previous year's sales [1]. Firms that have more growth opportunities tend to pay lower dividends and repurchase more shares. Thus, the expected relationship with the dependent variables is negative.

Crisis: it is a dummy variable that takes 1 for the covid 19 periods stretching from 2019 to 2020 to 2021.

MTB: Market-to-Book variable is defined by the ratio of the book value to the market value of equity [2].

Industry: The activity sector is a dummy variable that takes 1 for each Sector, and 0 otherwise.

A detailed description of all variables used in this study is presented in Table 3.

Table 3.

Variables measurements.

Variables (abbreviation)	Definitions	Measurements	Sources
DIVPAID	Dividend paid	Dividend payout per share	[1]
PERF	Financial performance	Return on assets (ROA) and return on equity (ROE).	[31]; Rime, 2001)
KZ	Financial constraints	KZ index	[44]
Indust	Activity sector	Dummy variable, which takes 1 for each Sector, and 0 otherwise	[2]
AGE	Firm age	Number of years since the company was founded.	[24]
Debt	Total debt	Ratio of total debt to total assets
Growth	Sales growth	Rate of change in sales scaled by the previous year's sales.	[1]
MTB	Market-To-Book	Ratio of the book value to the market value of equity.	[2]
Cashflows	Cash flows	Ratio cash flow to total assets.	[10]
Size	Firm size	The natural logarithm of total assets	DeAngelo et al. (2002)
Crisis	Crisis	A dummy variable that takes 1 for the covid 19 periods corresponding to 2019, 2020 and 2021.	[2]

3.3. Model and estimation method

To explore the relationship between dividend policy, performance, and financial constraints during the crisis period, we selected two models (eq (1) and eq (2)).

$${DIVPAID}_{it}={\alpha }_i+{\alpha }_1{DIVPAID}_{it-1}+{\alpha }_2{PREF}_{it}+{\alpha }_3{KZ}_{it}+{\alpha }_4{INDUST}_{it}+{\alpha }_5{AGE}_{it}+{\alpha }_6{DEBT}_{it}+{\alpha }_7{GROWTH}_{it}+{\alpha }_8{MTB}_{it}+{\alpha }_9{CASHFLOWS}_{it}+{\alpha }_{10}{CRISIS}_{it}+{\epsilon }_{it}$$ (1)

$${PERF}_{it}={\alpha }_i+{\alpha }_1{PERF}_{it-1}+{\alpha }_2{DIVPAID}_{it}+{\alpha }_3{KZ}_{it}+{\alpha }_4{INDUST}_{it}+{\alpha }_5{DEBT}_{it}+{\alpha }_6{GROWTH}_{it}+{\alpha }_7{CASHFLOWS}_{it}+{\alpha }_8{CRISIS}_{it}+{\epsilon }_{it}$$ (2)

${\alpha }_i$ represents the constant; $\alpha$ 1–10 are the coefficients of the independent variables; ${\epsilon }_{it}$ represents the estimation error; i denotes firms and t represents periods.

To check the causal relationships and the interrelationship between the variables, we formulated two dynamic regression models that took into account the effect of the lagged dependent variable. By including the previous value of the dependent variable, the model captures the adjustment process [46]. From an prior study, the past dividend and performance value has an impact on her current value. When applying a dynamic model, it is important to note that traditional estimation methods such as pooled ordinary least squares (POLS), fixed effects (FE), and random effects (RE) for panel data may not be suitable for this study, as they can introduce biases in the coefficient estimates. In this case, the most appropriate technique is system generalized methods of moments (SGMM). Moreover, This method is widely adopted in the finance literature (Attig et al., 2021). It is applied more in corporate financing decisions and in the interrelationship of financial decision-making such as investment, dividend, and external financing [47,48]. It has several advantages over the simultaneous equation system (SES) method. Indeed, it provides solutions to simultaneity bias, reverse causality, and omitted variables problems (Arrelando and Bond, 1991). Additionally, this approach resolves endogeneity biases of the independent variables and provides strong instruments for the endogenous lagged dependent variable. It utilizes the difference values of lagged dependent variables as instruments for the level equation and lagged dependent variables as instruments for the difference equation. This approach accounts for causal relationships and mitigates potential biases in coefficient estimations.

The validity of SGMM must meet three conditions such as:

• -Overidentification Restrictions: The overidentification restrictions, which are assessed through tests like the Sargan and Hansen tests, need to be satisfied. These tests evaluate the validity of the instruments used in the estimation process and the model's ability to capture the moment conditions accurately.
• -Absence of Autocorrelation: Autocorrelation, particularly of higher orders than 1, should not exist in the difference equation. This assumption is examined through tests like the Arellano-Bond (AR) tests.
• -Sufficient Number of Instruments: The number of groups, referring to the different entities in the dataset, should be greater than the number of instruments used. This condition ensures that there are enough observations to support the instruments' validity and to avoid the estimation becoming underdetermined.

4. Results

In this section, we first present the descriptive statistics of all variables used in the study. Second, we show their correlation matrix and finally, we report and interpret the empirical results.

4.1. Descriptive statistics

Descriptive statistics give an overview of the average value, minimum value, maximum value, and standard deviation of each variable. Table 4 gives the descriptive statistics of all variables used in this study during the crisis period, pre-crisis and the entire period.

Table 4.

Descriptive statistics.

	Crisis period					Pre-crisis period				Entire period
Variables	Obs	Mean	Std. Dev.	Min	Max	Mean	Std. Dev.	Min	Max	Obs	Mean	Std. Dev.	Min	Max
DivPaid	318	25.51	30.74	0	200	30.47	29.52	0	20.0	636	28.59	29.90	0	200
ROA	318	0.25	0.47	−2,05	3.32	0.49	0.45	−0.85	2.25	635	0.37	0.47	−0.05	3.32
ROE	318	0.47	1.15	−5,5	3.26	0.75	1.27	−13.1	3.36	636	0.611	1.22	−13.1	3.36
KZ	318	0.12	0.49	−0.3	5.75	−0.11	0.46	−0.16	5,24	637	0.11	0.48	−0.30	5.75
Age	318	70.27	47.08	17	316	70.72	47.08	17	316	627	78.02	51.18	19	316
Debt	318	1.31	9.57	0	133	0.46	1.26	0	11.8	636	0.88	6.83	0	5.75
Growth	318	1.54	7.57	−1	66.72	0.23	2.4	−0.28	42.73	636	0.88	5.65	−1	66.72
MTB	318	1.00	2.68	−18.2	12.26	1.6	6.28	−93.4	11.62	636	1.3	4.83	−93.4	12.26
Cashflows	318	1.05	0.45	0.5	1.6	1.0	0.66	0.18	1.85	636	1.02	0.573	0.18	1.85
Crisis	–	–	–	–	–	0	0	0	0	636	0.5	0.500	0	1
Size	318	15.73	1.93	9.93	23.36	15.53	1.86	9.89	19.20	636	15.63	1.9	9.89	23.36
Indust	402	6.91	2.64	1	9	6.91	2.64	1	9	636	6.93	2.57	1	9

During all period (2016–2021), the first dependent variable (Dividpaid) shows that the French companies paid dividends with a mean of 28.59. This dividend payout behavior seems however to be heterogeneous between firms-years with a standard deviation of 29.9.

Table 4 reports on dividend payout before and during Covid 19 crisis. There are more dividend reductions and fewer dividend increases which is consistent with the finding of Zhou and Ntantamis (2020) for the G7 countries. They showed that only France, Germany, Italy and the United Kingdom experienced a large dividend drop in 2020 compared to other countries in the group.

This highlights that managers prefer to opt for significant dividend cuts when they cannot avoid it. Moreover, this finding indicates that the firm's profitability (ROA and ROE) has experienced a great decrease: (49 % average for ROA before the crisis to 25 % during the crisis and ROE from 75 % to 47 %). However, despite this drop, the SBF 120 companies have higher profitability: average financial profitability during the entire period is 61.1 %. However, the KZ index and indebtedness have increased.

We continue our preliminary data analysis and examine the evolution of our main variables (dividPaid, Performance, and KZ index) during the entire period (Fig. 2). We notice that dividends decreased during the 2019–2020 period alongside with a decrease in the two performance measures (ROA and ROE). This confirms the descriptive statistics.

Fig. 2.

Variables evolution.

To deepen our results, it is essential to check multicollinearity problems between the variables. Multicollinearity can distort the accuracy of the regression estimation and make their estimated coefficients sensitive to small fluctuations in the data [49]. According to Brooks (2008), coefficients between −0.8 and + 0.8 indicate no multicollinearity problems. Table 5 shows the absence of multicollinearity this explained by the presence of lower correlation coefficients (less than 0.5).

Table 5.

Correlation matrix.

	DivPaid	ROA	ROE	KZ	Age	Indust	Debt	Growth	MTB	CASH FLOWS	Crisis	Size
DivPaid	1.00
ROA	0.053	1.00
ROE	0.091	0.037	1.0
KZ	−0.054	−0.09	−0.09	1.00
Age	0.077	0.11	0.06	−0.01	1.00
Indust	−0.098	−0.18	−0.06	−0.01	−0.09	1.0
Debt	−0.008	−0.05	−0.01	0.296	−0.06	−0.015	1.0
Growth	0.012	−0.02	−0.03	−0.07	0.056	−0.003	0.302	1.00
MTB	0.009	0.28	0.02	−0.01	0.010	−0.009	−0.02	−0.010	1.00
Cashflows	0.002	0.00	0.01	−0.48	−0.02	0.007	0.001	0.110	−0.02	1.00
Crisis	−0.075	−0.15	−0.06	0.009	0.004	0.173	0.062	0.118	−0.06	0.048	1.00
Size	0.28	0.08	0.04	−0.17	0.215	−0.072	−0.10	0.153	0.052	0.06	−0.03	1.00

This finding is also supported by the Variance Inflation Factor (VIF) test. VIF test is calculated for each model specification and the result show values lower than 2 (Table 6).

Table 6.

Vif test.

Variables	M1		M 2
	VIF	1/VIF	VIF	1/VIF
DivPaid	–	–	1.02	0.977937
ROA	1.14	0.878517	–	–
ROE	1.02	0.975932	–	–
KZ	1.54	0.651176	1.51	0.662200
Age	1.03	0.967741
Indust	1.05	0.956602	1.05	0.953770
Debt	1.27	1.27	1.27	0.786246
Growth	1.15	0.867231	1.15	0.868302
MTB	1.09	0.917475	–	–
Casflows	1.35	0.738394	–	–
Crisis	1.08	0.927264	1.06	0.944738
Size	1.02	0.979586	1.35	0.739609
Mean VIF	1.17		1.16

4.2. Empirical results and discussion

Table 7a, Table 7b presents the regressions estimation by SGMM method. The results report on the causal effects of performance and dividend policy of French companies during the health crisis. Three main tests were conducted in both models (Table 7a, Table 7ba and 7b): First, the Sargan test for the overidentification proves that instruments are valid and strong enough to deal with the endogenous problem caused by the lag of the dependent variable. Second The AR (1) and AR (2) tests indicate that the first-differenced errors have some first-order autocorrelation but not the second-order autocorrelation, which is acceptable in some cases. Finally, the number of groups is much higher than the number of instruments is a positive sign, as it suggests a rich cross-sectional dimension for identification.

Table 7a.

Results (dependent variable: DividPaid).

	Divpaid
	Coef.	Std. Err.	Z	P > z
DivPaidt-1	0.25692	0.06344	4.05	0.000
ROA	0.1162	0.083	1.40	1.63
ROE	0.2321	0.392	0.57	0.47
KZ	−0.00008	9.60e-06	−9.05	0.000
Indust	0.23599	0.42023	0.56	0.574
Age	0.0468	0.12268	0.38	0.703
Debt	0.06775	0.01661	4.08	0.000
Growth	0.18133	0.07669	2.36	0.018
MTB	−0.1261	0.09450	−1.34	0.182
Cashflows	0.16326	1.01259	0.16	0.872
Size	3.3307	1.61003	2.07	0.039
Crisis	−4.5370	1.33988	−3.39	0.001
Const	18.2457	9.70854	1.88	0.060
Observations	521
Number of instruments	23
Number of groups	105
Wald chi2	550.40
Prob > chi2	0.000
AR (1)	−4.5389
P-value	0.000
AR (2)	0.21309
P-value	0.8313
Sargan Test	13.4773
P-value	0.3353

Table 7b.

Results (dependent variables: ROA and ROE).

ROA					ROE
	Coef.	Std. Err.	Z	P > z	Coef.	Std. Err.	Z	P > z
ROAt-1	0.57817	0.05281	10.95	0.000
ROEt-1					0.75537	0.00969	77.93	0.000
DivPaid	−0.0221	0.011008	−2.01	0.044	−0.09132	0.05041	−1.81	0.07
KZ	−0.00002	2.35e-06	−9.97	0.000	−0.00001	4.24e-06	−3.59	0.000
Debt	−0.01006	0.018974	−0.53	0.596	−0.00722	0.00161	−4.49	0.000
size	0.19547	0.211291	0.93	0.355	7.3133	2.33389	3.13	0.002
Indust	0.04561	0.094061	0.48	0.628	0.01635	0.02164	0.76	0.450
Growth	0.041999	0.016178	2.60	0.009	0.00251	0.00597	0.42	0.674
Crisis	−1.3634	0.332293	−4.10	0.000	−0.18525	0.07147	−2.59	0.010
Const	0.93849	3.352101	−0.28	0.779	−0.88804	1.86082	−0.48	0.633
Observations	529				530
Number of instruments	22				22
Number of groups	106				106
Wald chi2 (8)	197.73				261.49
Prob > chi2	0.000				0.000
AR (1)	−3.6951				−2.002
P-value	0.0002 |				0.0453
AR (2)	−0.9526				0.6771
P-value	0.3408				0.4983
Sargan Test	15.8516				18.99655
P-value	0.2572				0.1232

First, we tested the DivPaid variable as a function of performance (ROA and ROE), the KZ index, and other control variables (Table 7a).

Table 7a shows that the variable (DivPaid) is positively significant at the 1 % threshold at DivPaidt-1 and negatively associated with the KZ index and Crisis at the 1 % threshold.

The covid19 crisis decreased dividend payout in the French context. This finding corroborates the results of [1] who found that the impact of covid 19 is smaller in European firms than in firms in North America and Japan.

The negative relationship between the KZ index and Dividpaid means that despite the constrained financial situation of French firms, they would continue to distribute dividends to satisfy stakeholders. This is justified by the positive significance of the constant variable (${\alpha }_i$) at the 10 % threshold. This finding also corroborates those on the dividend policy in times of crisis such as that of [2,15]. Hence, H3 is retained and H2 is rejected.

In fact, during uncertainty, bank financing becomes difficult. Firms will be forced to look forward to retaining a larger share of their profits rather than distributing dividends. They seek to improve their governance system and further protect investors during periods of financial instability by reinvesting their profits internally [15].

Both performance measures (ROA and ROE), did not affect the dividend distribution policy of SBF120 firms. This finding is consistent with those of [2] and contradict those of [15]. This contradiction is justified by the specificity of the contexts. This rejects H2. Notably, there is no uniform effect on dividends distributed during the COVID-19 crisis. Some firms chose to suspend or reduce dividends due to the economic and financial uncertainties caused by the pandemic, while others chose to maintain or even increase dividends. The decision to distribute dividends depends on many factors, which are unique to each company including the particular circumstances of each company, future economic prospects, regulatory requirements, shareholder expectations, and its ability to manage the challenges associated with the COVID-19 pandemic.

For the other variables explaining dividend payment (Table7a), we note that only the variables of size, debt, and growth positively correlate with the variable (DivPaid) at the 5 %, 1 %, and 5 % thresholds respectively. For the rest of control variables are insignificant.

The positive effect of the size variable (coef = 3.3307) means that large firms are characterized by capital dispersion and high agency costs. This requires intensive monitoring, control, and a high distribution policy to align interests and control agency costs. Notably, according to the agency and the signaling theories, large firms have fewer growth opportunities, which explains the high dividend payout. It indicates that larger firms are more likely to distribute dividends. We explain these results by the finding of [2] who stated that these large firms are more likely to use the free cash flow for distributing dividends rather than investing in new projects.

Moreover, the positive effect of debt on dividend payments means that managers take on debt to reward shareholders, despite financial difficulties as proposed by the signaling theory [36]. Debt can help firms cope with market disruptions by providing additional cash flow to fund operating activities and dividends. In times of crises, firms may find it more difficult to obtain funds through other means, such as equity offerings, which leads them to use debt to obtain needed funds. This result may be explained by the profitability and productivity of this debt, as French listed firms are already mature. This finding confirms the signal theory, which suggests that firms send positive signals to attract more investors by distributing cash dividends. However, it is important to note that debt can also have negative consequences for dividend payouts if companies cannot repay their debt due to financial difficulties.

The variable Growth has a positive and significant impact on dividend policy. These results affirm the finding of Hasan et al. (2021), who stated that firms tend to retain and reinvest the profits gained because they will be used as funds for further financing enlargements and growth.

Second, for the different determinants of performance (eq (2)), the results (Table 7b) show that both performance measures (ROA and ROE) are negatively associated with the DivPaid variable at the 5 % threshold. This means that in times of crisis when firms distribute dividends, cash reserves are depleted [16]. H1 is then confirmed. In particular, dividend distribution may weaken the firm's financial position and make it more difficult for it to cope with the crisis. On the other hand, investors may interpret dividend payouts as a sign of a lack of confidence in the firm's future performance.

For the relationship between the KZ index and performance (ROA and ROE), it is negatively significant at the 1 % level with low coefficients (−0023 % and −0015 %) in the two measures. Consequently, H4 is retained. Our finding is in line with the findings of [50,51]. In times of crisis, there was suspension or disruption of activities due to lockdowns, especially in 2020. Firms may face liquidity restrictions, higher borrowing costs, and uncertainties in financial markets, which may hinder their ability to invest in assets, resources, and growth opportunities. In addition, companies may also face solvency issues, which may cause them to reduce their capital investments, which may affect their long-term performance.

For the control variables, we note that debt has a negative effect on ROE at the 1 % threshold and no significant effect on ROA. These results are consistent with studies of [13,18]. These studies have shown that debt will influence the investment behavior and consequently, the financial performance of firms. In addition, in times of uncertainty, debt impairs the firm's ability to face economic and financial challenges and leads to lower financial performance.

Size also relates to financial performance, but positively at the 1 % threshold. This shows that large firms have sufficient financial resources to protect themselves against economic and financial disturbances. Therefore, the performance of largecompanies is higher than that of small companies. Thus, large firms are able to achieve more profit comparing to small companies [47].

As for the Indust variable we find a positive but not significant impact on both ROA and ROE. We conclude that the activity sector has no effect on firm's performance. A sector is not such important as the division of individual business units that generate income for the global economy.

Regarding the Growth variable we note a positive and significant effect on ROA with low coefficient (0.0419). While we find a positive but not significant impact on ROE. These findings are confirming with the work of [52] in the Croatian manufacturing industry, a high demand on firm's goodwill consequently increase sales and in the final analysis, lead to a better performance.

To support H5, we notice that the Crisis variable negatively affected performance for both measures at the 1 % threshold. It is −1.36 for ROA and −0.18 for ROE respectively. These results are consistent with those of [11,53]. Then, H5 is retained. Notably, because of the restrictions taken, the COVID-19 pandemic reduced revenues, increased costs, and generated financial difficulties for firms. On the other hand, some sectors benefited from this pandemic, such as e-commerce and healthcare services.

In summary, the COVID-19 crisis had a significant impact on the dividend distribution of French companies. As a result of the pandemic, many companies were forced to reduce or temporarily suspend their dividend payouts to cope with business and economic uncertainty. The decision bears on the overall financial health, future economic prospects, and contractual or legal obligations of the company.

This is most evident in hospitality and travel services, food services, and retail-related industries. In these sectors, many companies decided to cut dividends to support their short-term cash flow, which allowed them to face potential financial challenges and benefit long-term investors.

During the covid 19 crisis, dividends played a signaling role in the financial situation of the company. They are considered a sign of financial stability and assured investor loyalty [54]. By distributing dividends, companies can show that they can generate profits despite economic difficulties and uncertainty.

5. Conclusion, policy implications and limitations

This study examined the impact of the covid 19 crisis on dividend policy and performance while taking into account the effect of financial constraints in the French context. The SGMM was used to check this interplay.

The sample consisted of 106 SBF companies observed over 6 years. The study period was divided into 3 subperiods to assess the effect of the crisis: a pre-pandemic (2016–2018), the pandemic period (2019–2021), and all periods (2016–2021).

The empirical results showed, first, that only the Crisis and the KZ index negatively correlate with the level of dividend payment. Notably, during the covid 19 crisis, French companies did not take into account the performance of firms in the decision to pay profits to shareholders, as the relationship between the two performance measures ROA and ROE and DIVPAID is not significant. For the control variables, we found that only debt, growth, and size positively affected the dividend payout level.

Second, the results confirmed that the performance of French firms according to these two measures (ROA and ROE) depended negatively on dividpaid, the KZindex, and the crisis during the study period.

The findings are consistent with those of previous studies which showed that the Covid-19 crisis decreased dividend levels and performance, increased costs, and caused financial difficulties for firms [1,2,11,13,53].

The results have theoretical and managerial implications:

Theoretical implications: The study should enrich the literature on dividend distribution, performance, and financial constraints in times of crisis. For instance, the study tested and supported existing theories about the relationship between dividend payout and organizational performance, and showed how financial constraints affected dividend policies. It also explored the decision to distribute dividends and identified its determining factors in times of crisis.

Managerial implications: The findings of the study can provide valuable insights to managers to better understand the different consequences of their dividend distribution decisions. These latter can affect cash levels available to face the decline in activity and the additional costs associated with the crisis. The results of the study can also help identify, more clearly, decisions on dividend distribution taking into account company performance and financial constraints. Moreover, this study can help managers communicate with shareholders about dividend distribution decisions and their implications on the company's performance and financial constraints. In particular, the results identified the specificities of the French market, in particular dividends distribution, performance, and financial constraints.

Although this study found interesting positive insights into dividend policy, performance, and financial constraints in the French context, there are some limitations. The study is limited to French companies. We did not account for the specificity of other contexts and the results may be different from one country to another. In addition, in terms of selected variables, it will be better to introduce variables representing governance such as board, ownership structure, and committee indicators.

Data availability statement

Data will be made available on request.

CRediT authorship contribution statement

Saliha Theiri: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Salah Ben Hamad: Supervision, Conceptualization. Mouna Ben Amor: Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.