According to Assaf (2021), diagnosing the current position and producing results that serve as a basis for forecasting future trends is something done from the calculation of economic-financial indices extracted from the financial statements of companies. Allied with this, the author states that the temporal comparison of the indicators is indicative of the company's evolution and performance dynamically.

The most popular index among investors and managers is ROE (return on equity), as it portrays the efficiency with which the company employs shareholders' capital and what is its return. This measure is obtained by the ratio of net income to equity value (Higgins, 2014).

As for Catapan et al. (2011), ROE measures business profitability by analyzing the influence of profits on sales and efficiency in the use of assets. Kijewska (2016) states that this indicator measures the company's efficiency in generating returns from the resources invested by shareholders.

The DuPont method, developed by F. Donaldon Brown in the 1910s, decomposes ROE into three main elements: a) net margin, which is reflected in the company's commercial policies; b) asset turnover, which measures the degree of efficiency in the use of assets; c) financial leverage, which shows how much of the company's total assets are financed by third-party capital. Subsequently, two more items with a direct influence on ROE are incorporated into the model: financial cost and tax burden. This model is called the five-element DuPont model (Fernandes et al., 2014; Kijewska, 2016).

According to Bunea et al. (2019), the determinants of ROE are like levers available to financial managers to guide their decisions. This means that operational strategies and their economic performance - as well as financial performance - must serve as the basis for the actions of managers who aim to increase the return to the company's shareholders.

Additionally, to measure the impacts of elements external to the company on ROE, some studies include macroeconomic indicators in their models, (Anarfi et al., 2016; Ndlovu & Alegidede, 2015).

Table 1 shows the results of previous research on elements, internal and external, that impact ROE, such as profit margin, asset turnover, financial leverage, GDP, inflation, and basic interest rate.

Table 1
Previous studies on ROE´s determinants

Source: Developed by the authors.

It is observed that in all attributes analyzed, there are quite different results between the surveys. For example, financial leverage has a positive relationship with ROE in some studies (Bunea et al., 2019; Ghita-Mitrescu & Duhnea, 2017; Hsieh et al., 2020; Soares & Gaudi, 2011), while Anarfi et al. al. (2016), Burja and Marginean (2014), Carvalho et al. (2017), Kijewsa (2016), Manjunatha and Gujjar (2020), Manjunatha et al. (2018), Ndlovu and Alagidede (2015), Oriskoova and Paksiova (2018) and Weidman (2019) find a negative sign between financial leverage and ROE.

Research such as Kijewska (2016) and Soares and Gaudi (2011) conclude, in turn, that operating factors such as net margin and asset turnover explain more the behavior of ROE than the financial aspects themselves (financial leverage, for example ). Thus, actions taken by company managers that increase revenues and/or reduce costs and expenses tend to increase ROE. Added to this, it is observed that the company's form of financing, whether via equity or third-party capital, does not have a significant impact on ROE performance (Carvalho et al., 2017).

Liesz and Maranville (2008) state that, to increase ROE, the company must focus its actions on increasing operating profit, being more efficient in the use of its assets, optimizing the capital structure, reducing the cost of debt, and reducing the impact of the tax burden. Certainly, these decisions lead to different financial strategies.

Borrowed from the medical sciences, the term diagnosis means the study of changes in the expected behavior pattern in certain situations. In Administration, the analysis of performance indicators shows the effects of factors internal and external to the company on its economic and financial performance (Andekina & Rakhmetova, 2013).

The Covid-19 pandemic is an unprecedented event, exogenous to the company, but which affects its results. It impacts all economic agents, all businesses, and all countries; causes supply chain disruption, and represents a demand and supply shock (Salisu & Vo, 2020; Zheng, 2022).

As it is a recent event, all possible impacts on the financial performance of companies are not clear, especially those in the medium and long term. This requires financial managers and investors to make decisions in an unstable and unfamiliar environment, creating an even greater challenge. (Hassan et al., 2020; Salisu & Vo, 2020).

For Avelar et al. (2021), the impacts of the Covid-19 pandemic on the performance of companies should be measured not only by the price of their shares in the capital market but also by analyzing their financial statements. In addition to health actions, there is a need for measures to ensure the financial health of people, companies, and countries. The authors also emphasize the importance of understanding all the operational aspects of companies based on their economic and financial indicators. This provides support for managers' decisions related to resource allocation, maintenance of operations and competitiveness.

Fahlenbrach et al. (2021) conclude that the severe restriction on economic activities (due to social isolation) leads to reduced consumption, a drop in sales volume, and, consequently, a decrease in companies' profitability.

Shen et al. (2020) empirically show the negative and significant impact of the Covid-19 pandemic on the performance of publicly traded Chinese companies. According to the authors, restrictions on the movement of people and social isolation cause a drastic drop in demand, leading to a significant decrease in productivity, revenue, and, consequently, the ROE of companies.

The economic crisis reduces demand and, consequently, companies' sales. In the short term, cost containment does not follow the drop in revenues, negatively impacting business profitability. To measure the effectiveness of using assets to generate revenue, given these crisis scenarios, the inventory turnover, accounts receivable and accounts payable indicators are the most relevant, especially in the short term (Devi et al., 2020).

The research is justified by the importance that the ROE determinants maintain in this context since previous studies are inconclusive. In addition, it contributes with analyzes referring to Latin American countries and investigations of the impacts of Covid-19 on the results of companies.

The research hypotheses raised in this work are:

H1: The variables gross margin, sales growth, inventory turnover, average payment term, fixed asset turnover, and tax charges have a positive impact on companies' ROE.

H2: The variables selling and administrative expenses, average collection period, and debt ratio harm the ROE of companies.

H3: The Covid-19 pandemic negatively impacts the ROE of companies.

The research is based on secondary data collected directly from S&P Capital IQ, a subscription-based platform that provides business data and news and economic, financial, and commercial conditions for public and private companies around the world. The information used in this work comprises the economic and financial variables of non-financial and publicly traded companies. The annual data comes from the Income Statement and Balance Sheet financial reports disclosed and available on Capital IQ's own database. Macroeconomic information is extracted from the World Bank and the International Monetary Fund for each of the analyzed countries.

The sample, collected on June 15, 2022, comprises the three Latin American countries with the highest number of companies listed on the capital market: Brazil, Chile, and Mexico. These three countries represent 73% of publicly traded companies listed on Latin American stock exchanges. Table 2 details the sample distribution after eliminating companies that represent negative equity.

Data is collected annually. The period before the Covid-19 pandemic from 2017 to 2019 represents 1,224 observations, while the post-pandemic period, from 2020 to 2021, has 816.

Table 2
Sample distribution

Source: Developed by the authors.

The ROE is analyzed as a dependent variable of the model to identify its determining factors.

The analysis variables of this study represent the explanatory components of return on equity (ROE). These variables represent indicators related to factors internal to the company, segregated into operational and financial, as well as external elements, which are represented by macroeconomic indices.

Among the indicators related to the operation, the following are used: gross margin, a ratio of administrative and selling expenses to sales, inventory turnover, fixed asset turnover, tax charges, sales growth, average collection period, and average payment period.

Gross margin indicates how much of sales are available to pay fixed costs and expenses and increase profit. This allows for estimating the break-even point in sales (Higgins, 2014 p.42) and, similarly, shows that sales growth increases the company's profitability (Ullah et al., 2020) – both, therefore, positively impact the SWR.

The ratio of administrative and commercial expenses to sales, according to Caneghem et al. (2021), represents the ability to control expenses and efficiency in the use of resources by company managers. In other words, it is expected to harm ROE.

According to Berk and DeMarzo (2017 p. 79), the greater the inventory turnover, the fewer days of stopped goods, and the more efficient management of working capital, favoring ROE.

The average collection period informs how long sales are converted into cash, portraying the company's commercial policy. This means that the longer this period is, the lower the ROE. On the other hand, the average payment period expresses the moment when the company must have the resources to pay its suppliers. Thus, the longer this period is, the more breath the cash has - a positive effect on ROE (Berk & DeMarzo, 2017 p. 79; Higgins, 2014 p. 47).

The turnover of fixed assets represents the intensity of the use of capital or operational leverage in the company's activity. In other words, it signals the efficiency of using permanent assets to add value to the company (Higgins, 2014 p. 47).

Still, in the category of operational elements, Ullah et al. (2020) do not find a significant connection between tax charges and profitability, contrasting the studies by Soares and Gaudi (2011), which point to a positive relationship.

The indicator of a financial nature studied is indebtedness, which corresponds to the proportion between third-party capital and equity. Despite having an inverse relationship with ROE, its maximization is not necessarily positive. Therefore, the challenge for financial managers is to find a balance between the costs vis-à-vis the benefits of the level of indebtedness practiced (Endri et al. 2021; Higgins, 2014 p. 47; Nguyen et al., 2021; Ullah et al., 2020).

For each of the countries studied, selected macroeconomic indices are collected from the World Bank and International Monetary Fund databases. The gross domestic product (GDP) represents the annual growth of the economy in percentage. The inflation rate depicts the annual consumer price index also in percentage. Finally, the basic interest rate reflects the interest rate defined by the country's monetary policy. For Mucharreira and Antunes (2015), the increase in GDP leads to the country's economic expansion, positively impacting the profitability of companies. The interest rate effect can be positive or negative, depending on the company's capital structure.

It is important to point out that, as already mentioned, studies on the impact of these macroeconomic indicators on ROE are inconclusive. This means that there is no congruence between their effect on the profitability of companies (Anarfi et al., 2016; Barbosa & Nogueira, 2018; Kanwal & Nadeem, 2013; Mucharreira & Antunes, 2015; Ndlovu & Alagidede, 2015).

Table 3 details the model's explanatory variables with their calculation formulas, as well as the expected sign on the impact on ROE.

Table 3
Independent variable

Source: Developed by the authors.

To identify the behavior of key ROE factors - as well as the impact of the Covid-19 pandemic on the performance of Latin American companies - the data are examined with three-time frames. Following the example of studies such as Alsamhi (2022), Devid et al. (2020), and Krieger et al. (2021), three periods are considered: before the pandemic (from 2017 to 2019), after the pandemic (from 2020 to 2021) and the total period (from 2017 to 2021).

With the help of STATA® software version 14.2, statistical modeling is applied for the three periods separately. Subsequently, a comparative analysis is made between the results of these periods to highlight possible differences or similarities.

P Panel data, or stacked data, consider both the cross-section and the time series. This allows for analyzing the data considering both individual and time variations (Fávero, 2013). For Gujarati and Porter (2011) the estimation techniques that consider the heterogeneity of the data more efficiently measure the unobservable components. That is, they are more suitable for studying the dynamics of changes in cross-section.

Among the estimation techniques for panel data, the following stand out: a) the least squares model for stacked data, Pooled Least Squares (POLS), which estimates the regression ignoring the cross-sectional nature and time series of the data; b) fixed effects model (FEM) that combines the observations, but the heterogeneity of each individual (cross-sectional unit) is captured in the constant and time-invariant part; c) random effects model (MEA), which assumes that the intercept values are assigned randomly (Gujarati & Porter, 2011).

The choice of the most adequate model to be applied is based on the Chow, Hausman, and Breusch-Pagan tests. According to Fávero (2013), the Chow test allows for identifying among the POLS (H0) and fixed effects (H1) models the one that is more adherent. The Hausman test compares random effects (H0) and fixed effects (H1) models. Finally, the Breusch-Pagan LM test indicates which one is the most suitable between the POLS (H0) and randomly selected (H1) models. For each test, if the p-value is less than 0.10, the null hypothesis (H0) is rejected and the alternative hypothesis (H1) is accepted.

After applying the statistical tests, as shown in Table 4, the appropriate model to be used in this study is the fixed effects regression with panel data:

Table 4
Econometric model definition tests

Source: Developed by the authors.

Equation 1 demonstrates the econometric model and the variables used in this study:

where:

α = constant

β = variable coefficients

μ = error term

gmg = gross margin

sae = ratio of selling and administrative expenses to sales

sag = sales growth

fat = fixed asset turnover

inv = inventory turnover

dso = days sales outstanding

dpo = days payable outstanding

etr = tax burden

deb = debt ratio

gdp = growth domestic rate

cpi = consumer price inflation

jur = basic interest rate of the country

It is noteworthy that, additionally, the Breusch-Pagan and White tests are applied to analyze the heteroscedasticity of the model. The endogeneity test is also used considering the macroeconomic variables with a time lag of one year for each sample observation.

Table 5 presents the descriptive statistics of the dependent (roe), independent (gmg, sae, sag, fat, inv, dso, dpo, etr and deb) and control (gdp, cpi, jur) variables for the periods before the pandemic from Covid-19 (2017 to 2019), post-pandemic (2020 and 2021) and the total period analyzed (from 2017 to 2021).

Table 5
Descriptive statistics

Source: Developed by the authors.

The average ROE in the post-pandemic period compared to before the pandemic shows an increase of 1.8 p.p. (from 8.67% to 10.45%). For the total period, it appears that the ROE has an average of 9.38%, with a standard deviation of 13.24. The minimum value found is -28.58% and the maximum is 55.92%. This means that there is high dispersion and the data is heterogeneous.

Of the independent variables, it is observed that indebtedness, sales growth and average collection period are the ones that grew the most with 8.9 p.p. (from 80.2% to 89.1%), 3 p.p. (from 8.2% to 15.5%) and 3.4 p.p. (from 62.2% to 65.6%) of increase, respectively. On the other hand, the variables inventory turnover and tax charges suffer the biggest drops between the two periods, that is, they fall 5.1 p.p. (from 27.7% to 22.6%) and 2.1 p.p. (from 27.2% to 25.1%), in that order.

As for the macroeconomic indicators of the three Latin American countries analyzed (Brazil, Chile and Mexico), a decrease of 0.9 p.p. on average GDP. The average inflation rate shows an increase of 1.4 p.p. between the two periods. The base interest rate, however, does not oscillate. It is important to note that these variations are consistent with the impact of the Covid-19 pandemic on the countries' economies.

The correlation matrix between the variables expresses the behavior of the dependent variable concerning the independent and control variables. Tables 6, 7 and 8 demonstrate the correlation between the factors analyzed for the segregated periods: before the pandemic, from 2017 to 2019; after the pandemic, from 2020 to 2021; and the entire period studied, 2017 to 2021, respectively.

It appears that the gross margin, fixed assets turnover, inventory turnover, and debt ratio components have a weak or medium positive correlation with ROE and remain unchanged over the years. The indicators of administrative and commercial expenses on sales, average collection period, average payment period, and tax charges show a low and negative correlation with ROE. The highlight is the sales growth factor, which shows an increase in correlation with ROE from 0.15 to 0.40 between the moments before and after the Covid-19 pandemic.

It is worth noting that the three macroeconomic indices included in this study, GDP, inflation, and the basic interest rate, do not show a correlation with the ROE in the period before the pandemic, but they do show a positive, relevant correlation in the post-pandemic years.

Table 6
Correlation matrix – 2017 to 2021

Source: Developed by the authors.

Table 7
Correlation matrix – 2017 to 2019

Source: Developed by the authors.

Table 8
Correlation matrix - 2020 and 2021

Source: Developed by the authors.

The data obtained by the fixed effects regression model with panel data are presented in Table 9. With the analysis of the results in any period of time, it is verified that the factors, gross margin, sales growth and average term of payments present significance and positive effect on the ROE, corroborating the studies by Anarfi et al. (2016), Brunea et al. (2019), Hsieh et al. (2020) Of these, sales growth is the one with the highest positive correlation with ROE, especially in the post-Covid-19 pandemic period.

The tax burden indicator is also significant, but with a negative impact on ROE. These results are in line with the literature and previous studies (Beck & DeMarzo, 2017 p. 79; Nguyen et al., 2021; Ullah et al., 2020), suggesting that factors related to the sales operation have a fundamental impact on behavior of ROE, regardless of elements external to the company.

The factor selling and administrative expenses on sales, despite having a significance of 1% in the consolidated and pre-pandemic periods, becomes insignificant considering the period after the pandemic. On the other hand, the fixed asset turnover index indicates significance in the total and post-pandemic period, but it is not significant from 2017 to 2019. This result is in contrast to previous studies (Beck & DeMarzo, 2017 p. 79; Caneghem et al ., 2021) since they present different conclusions, depending on the moment analyzed. This implies that expense control is less relevant in times of crisis, such as the Covid-19 pandemic, than at other times. In contrast to the impact of the company's fixed structure, represented by the fixed asset turnover ratio, with greater weight in the company's profitability in periods of crisis.

The inventory turnover and average collection period indicators, on the other hand, show contradictory results when comparing the periods before and after the Covid-19 pandemic segregated with the total period (from 2017 to 2021). The inventory turnover ratio only expresses significance when looking at separate periods. On the other hand, the average term of receipts expresses significance only in the full period. These indicators are related to working capital management and from these results it can be inferred that inventory management is more relevant in the short or very short term, while the impact presented by the average collection period is more significant in the medium and long term deadlines.

The debt ratio, which represents the component of the financial decisions of the company, shows insignificance in any moment of time. Endri et al. (2021) and Nguyen et al. (2021) found a negative impact on ROE. It is worth noting that debt has a low correlation with ROE. This result suggests that decisions on how the company finances itself have no relevant impact on shareholder return, as long as debt interest does not cause perverse effects on the company's profitability.

Concerning macroeconomic factors, it should be noted that GDP and the basic interest rate are insignificant in any analyzed period, while inflation is highly significant when the entire period is considered. It is important to mention that macroeconomic indicators have a median correlation with ROE. This result is in line with previous studies (Anarfi et al., 2016; Barbosa & Nogueira, 2018; Mucharreira & Antunes, 2015), which found similar results. It can be deduced that effects external to the company are not relevant for the return to shareholders, mainly in the short term.

Table 9
Fixed effects regression model results

Source: Developed by the authors.Note: *, **, *** correspond to significance level of 10%, 5% and 1%, respectively, standard error in brackets, cons. represents alfa intercept of model regression line.

It is noteworthy that the results after applying White's robust correction do not change the observed results and conclusions. In other words, the macroeconomic variables remain non-significant and the other independent variables integrate the explanation of the ROE behavior. Furthermore, when adopting the time lag of the variables inflation, GDP and interest rate in one year, it is detected that they present a higher level of significance with the behavior of the ROE. This makes sense as these factors reflect consequences on the results of companies over an extended period.