Accounting and Non-Financial Information on Firms’ Profitability: Evidence from Greece and Cyprus

Abstract

This paper develops an evaluation of profitability for firms in Greece and Cyprus from 2005 to 2020. More specifically, it contains an investigation of comparative relevance and dominance of accounting versus non-financial variables, which affect the daily operations of firms, on the firms’ level of profitability. Moreover, this research examines the impact of corruption, unemployment, part-time employment and Research and Development (R&D) on the performance of companies, in order to help managers by giving them more information and assisting in long-term strategic planning. The results indicate that these variables do not have a large effect on the firm-level profitability of these two countries, which is largely influenced by profit margin and other interaction variables, such as profit margin on asset turnover ratio and equity multiplier. The findings underline that internal operational efficiency acts as the primary driver of short-term profitability, whereas macro-level indicators display weaker immediate associations. However, managing these structural elements remains strategically relevant for long-term springiness.

1. Introduction

A significant amount of economic literature is devoted to examining the factors that influence business profitability and efficiency. In particular, economists have found and examined a variety of groups of variables that affect business profitability (Spitsin et al., 2020). A crucial part of determining a company’s intrinsic value is projecting future profitability, but due to the number of factors influencing earnings performance, this process is still difficult (Artikis et al., 2024). Because of the high correlation between earnings and stock market performance, this topic is fundamental to accounting and finance study.

Profitability is essentially the most broadly applicable measure of economic efficiency, making it easier to compare earnings to utilized capital (Shvachych & Kholod, 2017). It is therefore a crucial indicator of financial success and a major factor in long-term sustainability, going beyond simply making money. In a capitalist system, maximizing profits is still the primary goal, but financial measures are now under more scrutiny. Stakeholder concern over corporate financial performance has grown in the modern environment, especially since the financial crisis (Mohamed Tailab, 2014).

As a result, profitability serves as a crucial indicator of an organization’s well-being and ability to provide value for stakeholders. Longevity requires constant profitability management and monitoring since well-informed decision-making allows a firm to adjust to changes in financial resources (Burja, 2011). From a theoretical perspective, the Resource-Based View (RBV) suggests that firms formulate distinct strategies based on their unique mix of resources and capabilities, thereby establishing sustainable competitive advantages (Barney, 1991; Penrose, 1959; Wernerfelt, 1984).

Even though a company’s profitability can be determined by financial measurements like sales, expenses and profit margins, non-financial factors are also very important in determining long-term financial performance.

Beyond financial indicators, non-financial1 variables play a decisive role in shaping a firm’s profitability and long-term sustainability. The literature argues that firms which incorporate innovation, sustainability, employee engagement and strong corporate governance structures exhibit greater stability and achieve superior long-term performance (Eccles et al., 2014; Gompers et al., 2003).

While developed markets have been widely studied, there is a gap in understanding how traditional accounting drivers interrelate with severe macroeconomic fluctuations in small open economies like Greece’s and Cyprus’s. This study uniquely compares internal accounting efficiency against external institutional stressors (corruption, unemployment) during a post-crisis period.

Greece and Cyprus were specifically selected due to their sharing strong cultural and social dimensions, yet they possess distinct economic structures (Greece being more diversified, Cyprus highly reliant on services/tourism). Furthermore, both underwent severe but structurally different economic shocks (a sovereign debt crisis in Greece vs. a banking bail-in in Cyprus), providing a unique comparative setting to test profitability drivers under extreme institutional stress.

The study focused on the impact of non-financial indicators on return on equity (ROE). Although ROE is a key measure of financial performance, it is based primarily on historical financial data. A more comprehensive assessment of corporate success requires the incorporation of non-financial factors (Kaplan & Norton, 1992). Specifically, the value and profitability of a firm are significantly influenced by intangible elements such as corporate culture, governance structures, corporate social responsibility (CSR) and employee satisfaction (Edmans, 2011; Friede et al., 2015). In addition, external and macroeconomic factors, such as investment in Research and Development, levels of corruption and labor market dynamics (e.g., unemployment and underemployment rates), are significant pillars that support firms’ growth and profitability (Eberhart et al., 2004; Mauro, 1995).

This research contributed to the relevant literature by analyzing combined accounting and non-financial factors that affect the profitability of firms in Greece and Cyprus. Its innovation lays in the application of the DuPont model in combination with macroeconomic and institutional variables (corruption, unemployment, R&D), something that has not previously been examined comparatively in the two countries. The study sought to answer the following questions: (a) To what extent do non-financial factors relate to profitability?; and (b) do these associations differ between the two countries? Summarizing, the results showed that accounting indicators have a statistically significant relationship, while non-financial variables showed a weaker correlation.

By bridging the gap between internal financial analysis, which is based on the Resource-Based View, and external systemic stress, which is based on Institutional Theory, this study significantly enhances the body of current literature. Internal operational efficiency serves as a main buffer against institutional volatility, as this integrated approach notably illustrates, by combining macro-institutional variables with the DuPont decomposition. As a result, it offers fresh perspectives on how businesses in open, tiny economies like Greece and Cyprus separate their profitability processes from sudden external macroeconomic shocks.

The paper is structured as follows: Section 2 presents the hypothesis of development and research design. Section 3 describes the data and methodology. Section 4 presents empirical results. Section 5 and Section 6 provide the concluding remarks.

2. Hypothesis Development and Research Design

2.1. Literature Review

The theoretical background of the study is based on the Resource-Based View (RBV) and Institutional Theory. The RBV explains profitability as a consequence of internal resources and skills, such as accounting performance indicators. In contrast, Institutional Theory emphasizes the influence of the external environment—institutions and political and social norms—on corporate behavior. The combination of the two approaches permits an interpretation of the differences between Greece and Cyprus.

Based on the Resource-Based View (RBV), internal accounting drivers (PM, ATO) are hypothesized to display a firm’s core competencies. Contrariwise, Institutional Theory posits that external macroeconomic variables (corruption, unemployment) exert pressure that can constrain profitability.

A firm’s resources that are valuable, rare, inimitable and non-substitutable (VRIN) produce sustainable competitive advantages, which in sequence can translate into higher profitability (Barney, 1991). Moreover, Peteraf (1993) identified four conditions needed (e.g., ex post/ex ante constraints, limited resource mobility) to drive a resource to a sustainable advantage and thus to superior financial performance.

Resources/capabilities such as IT capabilities are positively associated with firm performance, but are facilitated by organizational capabilities; that is, resources deliver profits only when they are combined properly. Going a step further, other research has shown that human capital and HR practices can be strategic resources with a direct impact on profitability, but only when these practices create hard-to-imitate capabilities and organizational synergy (Wright et al., 2001). Based on research regarding developments in the RBV area, there are shifts towards dynamic capabilities, resource systems and the link to financial indicators (such as profitability) (D’Oria et al., 2021).

Referring to Institutional Theory, focusing on profitability and financial performance, organizations adopt institutional norms/rituals to gain legitimacy; this often does not directly improve productive efficiency or profitability, but can facilitate access to resources (e.g., finance), so the indirect impact on profitability is significant (Meyer & Rowan, 1977). In addition to that, in periods of institutional change (e.g., transitional economies), strategic choices that fit the institutional environment (e.g., relationship with politics/local networks) lead to better performance. So “compatibility” with the institutional framework can increase profitability (Peng, 2003).

Institutional Theory proposes that in periods of economic pressure, external restrictions (such as high unemployment or corruption) create forcible isomorphic pressure, potentially counterbalancing firm-level strategies. This study tests whether core productivity (DuPont drivers) can overcome these institutional constraints.

As corporate social responsibility (CSP) practices are institutionalized in a field, the CSP–financial performance link changes over time. When practices become institutionally accepted, financial outcomes (e.g., profitability) may improve through legitimacy and access to resources (Brower & Dacin, 2020). Adding to this point, it can be mentioned that the implementation of “green” practices translates into financial performance when practices are supported by institutional legitimacy and internal capabilities—i.e., profitability requires both resources/capabilities (RBV) and institutional support (El-Garaihy et al., 2022).

Other studies showed a positive relationship between institutional quality (rule of law, regulatory quality, etc.) and financial performance of firms, which strongly indicated that the macro-institutional framework affects profitability (Eldomiaty et al., 2023).

Previous research in the field of non-financial reporting has indicated a significant development and the results of this path have attracted the interest of many stakeholders (Turzo et al., 2022). Non-financial information can assist the business in enhancing its future financial performance. Hristov et al. (2023) finds that there is a significant positive influence of stakeholders from non-financial resources on a firm’s profitability. Higher importance assigned to non-financial resources correlates with a higher return on sales.

Non-financial reporting about firms, such as those addressing ESG issues, has become increasingly important in recent years. Without considering how much emphasis a company has placed on ESG issues, it is impossible to evaluate the current potential for investment and, consequently, the value of the IT (Ali Khan & Obiosa, 2024).

Many studies have taken place around the world examining the implications of internal or external non-financial variables on firms’ profitability, such as corporate governance, corporate social responsibility (CSR) and ESG, innovation and R&D, industry, macroeconomic factors, the institutional and legal environment, and technological advancement.

Usman and Afandy (2022) examined the relationship between non-financial information and firm profitability in US-listed companies and found that non-financial information helps businesses establish credibility with stakeholders and improves their reputation, both of which may result in increased profitability in markets with fierce competition. Pedrini et al. (2023) investigated how stakeholders’ non-financial resources influence a firm’s profitability in Spanish companies and identify that they are broken down into key four value drivers that support different dimensions of a company’s performance. These drivers are organizational culture, employee motivation, organizational integration and stakeholder perception, which are found, both collectively and individually, to impact profitability, with a positive sign. Youssef et al. (2023) explored financial factors (profit margin, working capital management, liquidity, leverage) and non-financial factors (uncertainty, gross domestic product, inflation) influencing profitability in non-financial SMEs in the UK. They noted the importance of these determinants for managers and stakeholders in ensuring stability and sustainability.

Despite the various studies on the impact of non-financial variables on the profitability of firms in many countries, research in Southeastern Europe—and, more specifically, in Greece and Cyprus—is scarce. Attention has been paid to these two countries since financial indicator research on profitability in Greece and Cyprus may offer a new viewpoint. Cultural and social dimensions, the tourism-driven and service-oriented status of their economies, as well the population of SMEs in these two countries motivated us to focus on them.

This study investigates the relationship of accounting and non-financial variables on the profitability of firms in Greece and Cyprus over the period 2005–2020. The motivation for this study, as mentioned above, is a desire to conduct our research in these countries as a result of the similar cultural and social dimensions of the countries, the tourism-driven and service-oriented status of their economies, and the population of SMEs. At the same time, they have experienced periods of economic crisis, due to a debt crisis in Greece and a banking failure in Cyprus. Our basis will be the examination of ROE through a Dupont analysis. Non-financial variables also have been added, to claim the existence or non-existence of changes in the profitability of the companies in the two countries and what association they have with ROE.

2.2. Hypothesis Development

The necessary statistical tests have been conducted to see if an increase in the corruption variable would have any relationship with the companies’ ROE. In addition, this study indicates whether the existence of unemployment as well as part-time employment during the periods considered affect the profitability of the firms. Also assessed is whether the Research and Development costs of the two countries during the period 2005–2020 had a positive, negative or no particular impact on the examined firms in the two countries.

In addition, firms in these countries have been under pressure to enhance corporate governance and transparency, due to the debt crisis in Greece and the banking failure in Cyprus. A study in these countries could evaluate if increased profitability has resulted from improved disclosure of non-financial information. In this line, it must be stressed that investment decisions may increasingly be influenced by environmental, social and governance (ESG) factors (Park & Jae, 2021). In Greece and Cyprus, many businesses are adjusting to EU sustainability standards.2

Companies that put sustainability, ethical governance and employee welfare first are more competitive in global markets. Also, an investigation of successful Greek and Cypriot firms could demonstrate how non-financial factors affect profitability. By understanding the importance of non-financial elements, businesses in Greece and Cyprus can attract foreign investment. Moreover, firms with open corporate governance, moral business conduct and social responsibility programs are more likely to win over investors. In addition to the above, insights could help policymakers to improve economic stability and regulations. The results may help companies to enhance employee engagement, sustainability and corporate governance.

This study develops hypotheses to understand how and at which level non-financial information acts with respect to firm profitability in both countries. Attention has been paid to variables that may play an important role in the sustainability and evolution of businesses. An analysis was conducted to determine whether corruption levels affect business profitability. Corruption is a pervasive institutional factor in a country, permeating business behavior and influencing corporate misconduct. Moreover, the higher the level of corruption in the country, the smaller the contribution of capitalized development expenditure each year to future profitability (Mazzi et al., 2019). Being a permeable and informal country characteristic, corruption is pervasive in business activities and dealings (Rodriguez et al., 2005), with negative consequences. Corruption may recede as firms become more exposed to international rules (Reid, 1983).

Policymakers can develop ways to strengthen institutional frameworks by recognizing the complex link between corruption, corporate profitability and the function of auditors (Sundarasen et al., 2024). In some contexts, corruption has the ability to generate benefits or opportunities that improve performance momentarily, but in other contexts, it can have negative results, such as reduced competitiveness, increased legal risk and reputational damage. These impacts would hurt an organization’s profitability and competitiveness.

The study’s hypotheses are formulated using two complementary theoretical approaches: the Resource-Based View (RBV) and Institutional Theory.

According to the RBV (Barney, 1991; Wernerfelt, 1984), a firm’s internal resources and capabilities—such as accounting performance indicators and efficient capital utilization—are sources of sustainable competitive advantage and, consequently, higher profitability.

At the same time, Institutional Theory (DiMaggio & Powell, 1983; North, 1990) emphasizes that corporate performance is determined not only by internal characteristics, but also by macroeconomic and institutional conditions in the operating environment, such as corruption levels, unemployment and R&D investment.

From an Institutional Theory standpoint, high corruption creates institutional cavities that increase transaction costs and sidetrack resources away from productive activities, thereby theorized to negatively impact firm-level profitability.

The integration of the two theoretical approaches provides a comprehensive framework for understanding the factors that influence profitability, allowing for the simultaneous investigation of accounting (internal) and non-financial (external) variables.

According to Institutional Theory, excessive levels of corruption lead to institutional gaps that raise transaction costs, take funds away from profitable endeavors, and increase threats to one’s reputation and legal standing. It is hypothesized that these inefficiencies will lower overall business performance and decrease profit margins.

Consequently, the following is the formulation of the first hypothesis.

H1. 

Firm profitability is inversely correlated with corruption.

This research also examined how much business growth in a larger economy is hampered by job loss or the transition from full-time employment to unemployment and/or part-time employment. This study specifically investigated whether underemployment and/or part-time employment cause a decline in consumer demand, investment capacity and labor productivity, hence a reduction in the likelihood of long-term economic growth. The implications of science and technology are considered since, in a world that keeps getting more digitalized, companies must adjust and enhance their approach to the new environment. The way that work is organized and performed, the skills that are required to do the work, the employment relationships, the social protection system, the formalization of informal sectors and job quality have significantly been altered (Charles et al., 2022). One of the biggest factors leading to business bankruptcy in Lithuania is unemployment (Bekeris, 2012). A high unemployment rate puts pressure on corporate revenues and profitability by lowering the general demand for goods and services (Kroft et al., 2016).

Long-term unemployment lowers the productivity and quality of work produced by available workers, which has a detrimental association with profitability through decreased performance and increased expenditures for training and replacement (Abraham et al., 2019; Benigno et al., 2015). Additionally, part-time plans may result in employees investing less in human capital and being less committed, which over time hinders innovation and earnings (Devicienti et al., 2018; Künn-Nelen et al., 2013). A large percentage of part-time employment without adequate planning might lower total productivity (TFP), particularly in horizontal occupations that require coordination and skill dispersion (Devicienti et al., 2018). Part-time employment and unemployment have complicated, two-pronged effects on profitability: they can lower expenses, but they can also lower productivity and innovation. The ultimate result is contingent upon industry, management and framework regulations (Künn-Nelen et al., 2013; Devicienti et al., 2018; Abraham et al., 2019; Benigno et al., 2015).

Taking the above into consideration, the second hypothesis has been developed: that unemployment and part–time employment have a negative relationship with profitability.

H2. 

Unemployment and part-time employment have a negative association with profitability.

AI will not necessarily lead to the skyrocketing unemployment rates that prior literature has suggested, because labor markets are already adapting to the new wave of technological change. Workers are responding by changing their occupations or by becoming entrepreneurs (Fossen & Sorgner, 2021). Since R&D expenditure creates new goods and markets, it has a long-term positive correlation with profitability (Khan, 2023). Additionally, due to product diversification, businesses that continuously spend on R&D are more profitable during difficult times (Sulimany, 2025). Moreover, universities, clusters and joint ventures working together on R&D improve return on investment, cutting expenses and raising profits (Yang, 2010). According to Leung (2021), in knowledge-intensive industries, R&D has a beneficial impact on productivity (TFP) and, consequently, profitability. In traditional industries, the impact of R&D is smaller, but it can increase profitability through process improvement (Hsu, 2020).

In general, R&D has a favorable but ambiguous impact: it increases risk and diversifies results by industry while providing significant long-term profits (Curtis et al., 2016).

Taking the above discussion into consideration, the third hypothesis has been developed: that R&D has a positive relationship with profitability.

H3. 

Research and Development has a positive association with profitability.

These assumptions are in line with the RBV, since accounting ratios reflect the internal capabilities of a firm to create and maintain profitability. At the same time, the inclusion of macroeconomic and institutional variables is based on Institutional Theory, which explains that external factors can limit or enhance the utilization of internal resources. Thus, hypotheses H1–H3 examined the composition of these forces in the context of the economies of Greece and Cyprus.

3. Data and Methodology

3.1. Data and Sample Description

To address the above research questions, financial information from 263 firms (

Table 1. Industry distribution.

Industry	Cyprus	Greece
	Number of Firms
Banks	1	6
Basic Resources	1	17
Chemicals	1	1
Construction and Mats	4	16
Consumer Prod and Svs	1	13
Drug and Grocery Stores	1	6
Energy	4	4
Financial Services	6	4
Food, Bev. and Tobacco	9	19
Health Care		9
Ind. Goods and Services	4	28
Insurance	4	2
Media		6
Real Estate	10	11
Retailers	3	10
Technology	5	15
Telecommunications		5
Travel and Leisure	20	9
Utilities		8
Total	74	189

) in both countries is gathered. The period examined ranges from 2005 to 2020.

The analysis is based on a sample of 263 listed firms. This study collected data from 74 firms in Cyprus and 189 firms in Greece. Their data were collected from the Datastream database for the period 2005–2020, based on the completeness of their available financial data. The sample also included firms from the financial sector due to their significant share of the two economies and their systemic importance.

The difference in sample size between Greece (n = 189) and Cyprus (n = 74) directly reflects the actual composition and size of the Athens Stock Exchange compared to the Cyprus Stock Exchange. Artificially matching the sample sizes would necessitate skipping a significant portion of the Greek market, leading to a critical loss of representative data and introducing survivorship or selection bias.

This timeframe was selected to depict a complete economic cycle, embodying the pre-crisis peak, the sovereign debt and banking crises and the subsequent recovery. Data after 2020 were purposefully excluded because the COVID-19 pandemic—and the unprecedented state subsidies that followed—introduced major structural breaks that would have significantly distorted the baseline profitability trends.

The financial sector is encompassed due to its systemic prominence in the Greek and Cypriot economies, especially during the crisis period. Not including it would cause a significant loss of evidence regarding the total market reaction to macro-institutional shocks.

The non-financial variables are collected from world values surveys, IMF datasheets, Eurostat, World Development Indicators, Bank of Greece, Trading Economics, Transparency International, Hellenic Statistical Authority, and Cyprus Statistical Service.

Taking a closer look, the results indicate that the Greek economy is distinguished by a broader and more developed industrial sector, with a much larger representation in the sectors of basic resources, construction, consumer goods, technology, retail, health, media and telecommunications. In contrast, the Cypriot economy appears to be more reliant on the travel and leisure sector, while maintaining a sizeable presence in real estate and financial services. While there are sectors with similar or low activity in both countries, such as banks, pharmaceuticals, chemicals and utilities (in Cyprus), the significant differences in the distribution of companies by sector underlines the structural specificities and different growth paths of the two economies.

Table 2. No. of observations per year according to country.

Country-Year	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Cyprus	46	51	51	52	54	48	52	53	57	55	56	53	58	52	52	51
Greece	154	163	164	164	168	166	166	171	172	171	167	167	163	161	159	150
Total	200	214	215	216	222	214	218	224	229	226	223	220	221	213	211	201

, presented below, shows the number of observations per year according to country.

Important DuPont analysis components are computed. With DuPont analysis, it is possible to identify which factors most affect a firm’s ROE, as well as the weaknesses and strengths of a business.

Next, the study used DuPont analysis formulas to determine ROE. Lastly, to identify the main elements influencing a company’s ROE, data has been examined.

Table 3. Descriptive statistics.

Panel A: Accounting Variables (DuPont components)—No. of Observations: 3356
Variable	Total obs.	mean	min	25%	50%	75%	max	Std. Dev.
ROE	3356	−0.01	−0.32	−0.04	0.00	0.03	0.21	0.08
PM	3356	−0.11	−5.24	−0.09	0.01	0.06	1.04	0.69
ATO	3356	0.69	0.01	0.27	0.51	0.84	5.68	0.82
EQM	3356	26.17	0.73	4.03	7.34	16.60	629.26	77.25
PM × ATO	3356	0.00	−0.26	−0.03	0.00	0.03	0.24	0.07
Panel B: Non-financial Variables—No. of Observations: 3467
Variable	Total obs.	mean	min	25%	50%	75%	max	Std. Dev.
CORR	3467	0.47	0.34	0.43	0.46	0.50	0.66	0.09
PTE	3467	0.08	0.04	0.06	0.09	0.10	0.13	0.02
UNP	3467	0.15	0.04	0.09	0.16	0.21	0.28	0.07
RD	3467	0.01	0.00	0.01	0.01	0.01	0.02	0.00

Note: Accounting variables have been winsorized at the 1–99% level. Sample included Greece and Cyprus and covers the period from 2005 to 2020.

presents the descriptive statistics for the sample. The descriptive analysis of the accounting variables showed a negative average for profitability (ROE) and profit margin (PM), which indicates overall profitability challenges in the observed group. Thus, based on all observations and the average price, it can be concluded that a company is not profitable. There is a structural sustainability problem if the situation persists. Shareholders are losing value and are not simply failing to profit from the companies in which they invest. Companies are not attractive to investors or creditors unless there is a clear recovery plan.

Although the turnover rate (ATO) presents a moderate average, the considerable standard deviation suggests significant variability in the efficient use of resources. Notably, the high mean and variability in the equity multiplier (EQM) suggest a frequent dependence on financial resources. The near-zero averages of the product of margin of equity and turnover (PM × ATO) further intensified concerns about profitability. Regarding the non-financial variables (CORR, PTE, UNP, RD), both their respective means and standard deviations indicated different levels of central tendency and dispersion. The accounting variables have been winsorized at the 1–99% level to mitigate possible outliers.

To diminish the influence of “risky” outliers, which are common in accounting data, a winsorization was applied at the first and 99th percentiles, safeguarding that the results are driven by the general population of firms rather than irregular tail behavior.

While traditional macroeconomic variables (such as inflation or exchange rates) depict cyclical economic fluctuations, this study purposely selected corruption, R&D expenditure, unemployment, and part-time employment because they operate as critical country-level proxies for the broader institutional and structural environment. Specifically, these pointers reflect institutional voids, the national innovation ecosystem, and labor market flexibility. Although their statistical association with short-term firm profitability may appear weak compared to internal accounting drivers, their inclusion is pivotal; it allows for a direct empirical test of Institutional Theory against the Resource-Based View within the same modeling framework.

3.2. Methodology

3.2.1. Original DuPont Model

$$\mathrm{R}\mathrm{O}\mathrm{E}=\mathrm{P}\mathrm{M}\ast \mathrm{A}\mathrm{T}\mathrm{O}\ast \mathrm{E}\mathrm{Q}\mathrm{M}$$

It is admitted that PM is algebraically linked to ROE. However, following the framework of Barbier (2020) and Soliman (2008), regression analysis is proposed to decompose the variation in ROE and to quantify the relative explanatory power of accounting drivers versus non-financial variables, rather than establishing a strict causal inference.

As Barbier (2020) demonstrated, the use of regression in the DuPont model is not to confirm mathematical equality (which is a given), but to measure the statistical significance and beta coefficients of each factor in determining ROE in a specific sample of companies.

In

Table 4. Definitions of variables.

Variable	Measurement	Source
ROE	Return on Equity = Net Profit/Average Equity	Nissim and Penman (2001) Barbier (2020) Penman (2013)
PM	Profit Margin = Net Profit/Average Net Sales	Fairfield and Yohn (2001) Nissim and Penman (2001) Wahlen et al. (2023) Nissim and Penman (2001) Barbier (2020)
ATO	Asset Turnover = Net Sales/Average Total Assets	Soliman (2008). Nissim and Penman (2001). Barbier (2020).
EQM	Equity Multiplier = Total Assets/Average Equity	Soliman (2008). Nissim and Penman (2001). Riahi-Belkaoui (2003). Barbier (2020)
PM × ATO	Interaction term between Profit Margin and Asset Turnover (PM × ATO)	Soliman (2008). Fairfield and Yohn (2001). Nissim and Penman (2001). It is used to examine the combined effect of operational efficiency and sales effectiveness.
CORR	Corruption Index, scaled between 0 and 1 (Transparency International data)	Mauro (1995). Dreher and Schneider (2010) Hoinaru et al. (2020)
RD	Research and Development expenditure as a % of GDP (scaled: value/100)	Artz et al. (2010). OECD (2024).
PTE	Part-Time Employment as a % of total employment (scaled: value/100)	Eurostat (2024). Garnero (2020).
UNP	National Unemployment Rate (scaled: value/100)	Blanchard and Wolfers (2000) Nickell et al. (2005) World Bank (2024)

below, the definitions of the variables which are used in the methodology are presented.

The definitions and measurements of the variables follow established approaches in the international literature (Nissim & Penman, 2001; Soliman, 2008; Hoinaru et al., 2020) to ensure the comparability and validity of the results.

Decomposition of ROE to a forecasting context has been analyzed—similarly to (Nissim & Penman, 2001; Barbier, 2020; Soliman, 2008)—as follows:

$${\mathrm{R}\mathrm{O}\mathrm{E}}_{it+1}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1{\mathrm{R}\mathrm{O}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4\left({\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}\right)+{\mathsf{\beta }}_5{\mathrm{E}\mathrm{Q}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}\mathrm{t}+1}$$

(1)

This study extended Equation (1) by adding four further variables, namely CORR (corruption), RDEXP (gross domestic expenditure on R&D as a percentage of GDP), part-time employment (PTE) and unemployment (UNP). Therefore, the following expression results:

$$\begin{array}{c}{\mathrm{R}\mathrm{O}\mathrm{E}}_{\mathrm{i}\mathrm{t}+1}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1{\mathrm{R}\mathrm{O}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4\left({\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}\right)+{\mathsf{\beta }}_5{\mathrm{E}\mathrm{Q}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{C}\mathrm{O}\mathrm{R}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_7{\mathrm{R}\mathrm{D}\mathrm{E}\mathrm{X}\mathrm{P}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_8{\mathrm{P}\mathrm{T}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\\ {\mathsf{\beta }}_9{\mathrm{U}\mathrm{N}\mathrm{P}}_{\mathrm{i}\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}\mathrm{t}+1}.\end{array}$$

(2)

Our approach goes beyond static ratio analysis to investigate the persistence and forecasting relevance of profitability drivers, in line with the strategic perspective paradigm proposed by Wahlen et al. (2023). Our regression model (Equation (2)) uses a dynamic lead–lag structure in which independent variables at time t predict return on equity at time t + 1 ($ROE\_it+1$), whilst the conventional DuPont identity reflects present performance.

Two factors make this temporal distinction crucial. Firstly, it is consistent with the valuation goal of financial statement analysis, which holds that non-financial elements and current financial ratios (PM, ATO) are important only to the degree that they predict future economic advantages. The study examines the durability of profit margins and asset efficiency as long-term sources of competitive advantage by regressing $ROE\_it+1$ on lagged components. Secondly, the model explicitly incorporates the trade-off between profitability and efficiency through the interaction term ($PM\_it x ATO\_it)$. According to Wahlen et al. (2023), this interaction captures the firm’s strategic positioning—distinguishing between differentiation strategies (driven by high margins) and cost leadership strategies (driven by high turnover).

Therefore, Equation (2) evaluates not just the individual contribution of accounting and non-financial variables (CORR, RD, PTE, UNP), but their combined capacity to determine the firm’s future financial trajectory in the subsequent period (t + 1).

The regression specification did not aim to establish causal relationships but to decompose and compare the relative explanatory contribution of accounting components and non-financial components embedded in the DuPont identity.

Through regression, this paper also followed Barbier’s methodology to identify which of the sub-indicators (e.g., operating margin vs. leverage) and/or non-financial variables have the greatest impact on ROE volatility. The regression framework computes the relative statistical significance and explanatory power of the mathematical components of ROE, whereas DuPont analysis finds them. The methodological approach is based on the framework of Barbier (2020) and Nissim and Penman (2001) too, who use econometric models (MLR) to isolate the effects of operational and financial activities on return on equity.

Due to the minor number of firms in specific sectors (e.g., in Cyprus, some sectors have only 1–2 firms), including sector fixed effects would harshly reduce degrees of freedom and lead to multicollinearity. Thus, there is the control for heterogeneity via the System GMM estimator.

3.2.2. Estimation Method

To control unobserved heterogeneity, potential endogeneity of explanatory variables and dynamic panel bias, the System Generalized Method of Moments (System GMM) was employed, an estimator proposed by Arellano and Bover (1995) and Blundell and Bond (1998).

The following dynamic panel model has been estimated:

$$y_{it}=a{y}_{it-1}+\beta X_{it}+{\eta }_{it}+u_{it}$$

(3)

where:

• $y_{it}$ is the dependent variable;
• $y_{it-1}$ is the lag of the dependent variable;
• $X_{it}$ represents the explanatory variables;
• ${\eta }_{it}$ represents the unobserved firm-specific effects;
• ${\epsilon }_{it}$ is the error term.

And the first-differenced equation is specified as follows:

$$∆y_{it}=∆a{y}_{it-1}+∆\beta X_{it}+∆u_{it}$$

(4)

In the GMM calculation of future ROE (2), PM has been treated as a fixed variable, even though the DuPont decomposition defines ROE algebraically as the product of PM, ATO and EQM. It is presumed that PM at time t is established before the error term ut + 1 is realized and so meets the orthogonality requirement E[PM t·ut + 1] = 0. Although the definitional connection between ROE and PM raised a stringent endogeneity worry, it is contended that the timing of information flows, in which PM is noticed before ROE realization, permitted it to be regarded as exogenous in this situation. Furthermore, to ensure the reliability of our inference, robust standard errors have been reported. This adjustment corrects for the presence of heteroskedasticity and serial correlation in the error structure, which are common in panel data analysis of corporate performance.

In addition to the use of GMM, the regressions with least squares have been estimated, separating the data of the two countries for Greece (Table 6, Appendix A and Appendix B) and Cyprus (Table 10, Appendix A and Appendix B).

The main regression model is:

$${\mathrm{R}\mathrm{O}\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1{\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\left({\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}\right)+{\mathsf{\beta }}_4{\mathrm{E}\mathrm{Q}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{C}\mathrm{O}\mathrm{R}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{R}\mathrm{D}\mathrm{E}\mathrm{X}\mathrm{P}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_7{\mathrm{P}\mathrm{T}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_8{\mathrm{U}\mathrm{N}\mathrm{P}}_{\mathrm{i}\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}\mathrm{t}}.$$

(5)

According to OLS, all the examinations for all variables and their coefficients have taken place. The statistical significance of each variable was tested and presented the effect results on the dependent variable, ROE. Also, the necessary tests for autocorrelation were done, as well as heteroskedasticity and multicollinearity, and the study went further to depict a safe conclusion.

The period 2005–2020 was chosen because it covered critical phases of the two economies, including the global financial crisis and the debt crisis. Time dummies were not included in the dynamic GMM model for two reasons. Firstly, the sample covered a relatively limited time period (2005–2020) with annual observations, which would lead to overparameterization and loss of degrees of freedom. Secondly, the model already included macroeconomic and institutional variables (corruption, unemployment, part-time employment, R&D) that reflected temporal variations in the economic environment, thus reducing the need for additional dummies. Macroeconomic variables are included to capture cross-country institutional variation rather than time-series shocks. Year fixed effects are intentionally omitted to avoid absorbing the explanatory power of slowly moving institutional indicators.

As Arellano and Bover (1995) and Roodman (2009) pointed out, in samples with a short time depth, the addition of time constants can cause estimation instability and increase the endogenous variability of the instruments, without substantially improving the explanatory power of the model.

4. Empirical Results

4.1. System GMM Results

The results indicate a statistically significant positive effect of profit margin (PM) and asset turnover (ATO) on ROE, confirming the RBV perspective of internal efficiency.

The statistical significance of profit margin (PM) indicates that firms implementing effective cost management strategies and maintaining pricing power demonstrate superior returns. Also, the positive coefficient of asset turnover (ATO) confirms that firms capable of generating higher revenue per unit of asset are more profitable, aligning with efficiency theories.

According to Shafiee et al. (2012) firms should concentrate on their margins and how effectively they are using their resources since that is what drives their earnings. This approach aligns with the principle that understanding and managing margins through efficient resource use is crucial for profitability.

Table 5. GMM estimation results—no. of obs.: 2.750.

Variable	Coef.	Std. Error	z-Stat	p-Value
L1.ROE	0.0206 *	0.0107	1.93	0.054
PM	0.0076 ***	0.0029	2.61	0.0089
ATO	0.0179 ***	0.0057	3.16	0.0016
EQM	−0.000004	0.000013	−0.27	0.784
PM × ATO	0.9456 ***	0.0233	40.60	0.000
CORR	−0.0055	0.0079	−0.69	0.491
RD	−0.1458	0.1754	−0.83	0.406
PTE	0.0359	0.0305	1.18	0.239
UNP	0.0039	0.0078	0.50	0.617
constant	−0.0136 ***	0.0034	−4.03	0.0001

Note: * p < 0.1; ** p < 0.05; *** p < 0.01.

allows us to discuss the aforementioned findings, as follows. Although there was awareness that an issue like corruption can cause problems for companies, it did not appear to be a significant factor that directly affects their return on equity (ROE). It is possible that businesses have developed adaptation strategies for how to handle it, or perhaps most of the organizations in the research study are dealing with it at comparable levels.

The results of the analysis showed that both profit margin (PM) and the asset turnover ratio (ATO) had a positive and statistically significant effect on return on equity (ROE), with p values of 0.009 and 0.002 respectively. This means that the more profitable a firm is and the more efficiently it uses its assets, the higher the return on equity. Particularly significant is the interaction between PM and ATO (PM × ATO), which is positive and strongly statistically significant (p < 0.001), indicating that the combination of profitability and asset utilization efficiency played a crucial role in improving ROE. In contrast, the financial leverage ratio (EQM) and the variables CORR, RD, PTE and UNP did not show statistically significant effects (p > 0.10), while the model constant is negative and statistically significant (p < 0.001), which may suggest that, in the absence of the effects of the independent variables, the expected value of ROE is low or negative (Table 5).

To frontier instrument proliferation, the ‘collapse’ option in the GMM estimator has been employed. The Hansen test p-value (0.119) approves the validity of the instruments used.

The Hansen test for overidentifying restrictions gave a chi-squared statistic of 123.354 with 106 degrees of freedom and a p-value of 0.119. This p-value is above the typical significance levels, meaning that the study did not reject the idea that the instruments used in the model are valid—they did not seem to be correlated with the error term, which is a good sign.

Moving on to the Arellano–Bond test for autocorrelation, a significant first-order autocorrelation in the different residuals (AR (1) p-value = 0.000) was seen, which is often expected in GMM estimation. However, the test did not find evidence of second-order autocorrelation (AR (2) p-value = 0.625), which is what was expected for a well-specified model.

Interestingly, taking on more debt (leverage) did not seem to be helping companies make more money right now. This trend may indicate a cautious approach to borrowing after the crises, or maybe it is just harder to get loans.

In addition to the above comments, how many people are unemployed in the general economy did not seem to directly impact how profitable individual companies are. This lack of statistical significance implies that, either because of adaptive staffing tactics or the prevalence of internal cost control, firm-level profitability in the sample studied may be shielded from more general labor market volatility. However, just using more part-time workers does not automatically mean a company will make more money. It could be that these flexible arrangements are not being used as effectively as they could be.

Finally, it looks like investing in new ideas and research (R&D) is not showing big profit boosts right away. Maybe these kinds of investments take a while to pay off, or maybe not enough companies are doing it, or the money is not always going to the right places to see a quick impact on the bottom line. It could also be that only a few companies are really investing heavily in R&D, so it is not a big effect across all the companies which have been looked at.

In conclusion, the findings suggest that to make a good profit these days, especially with how things are, companies really need to be sharp about managing their spending and selling their stuff well. They have got to handle tough situations. Things that might be assumed to be big factors, like how much debt they have or the general unemployment rate, did not seem to be directly shaking up their profits at the moment. Also, Research and Development approaches might take a bit of time to turn into bigger earnings.

4.2. OLS Results

4.2.1. Greece

Analyzing the Greek companies, based on the OLS results, the following results are obtained. The observations consist of 2555 entries and concern 189 firms for the period under consideration, 2005–2020.

According to general regression statistics for Greece using Ordinary Least Squares (OLS), the R-squared of 0.961 showed that the 96.1% of the variation in return on equity (ROE) is captured by this regression model, pointing to an exceptionally strong fit. The adjusted R-squared, also at 0.961, being virtually identical to the R-squared, suggests that the included independent variables are meaningfully contributing to explaining ROE without causing overfitting. Further bolstering this, the very high F-statistic of 7784.0 with a corresponding p-value of 0.00 strongly indicated that the model as a whole was statistically significant, meaning at least one of the independent variables had a real impact on ROE.

Table 6. OLS with respect to Greece, coefficients and Significance Test Statistics—no. of. obs.: 2555.

Variable	Coefficient	Std. Error (OLS)	t-Value	p-Value	95% Confidence Interval
const	−0.0010	0.003	−0.285	0.776	(−0.008, 0.006)
PM	0.0053 ***	0.001	9.234	0.000	(0.004, 0.006)
ATO	−0.0002	0.000	−0.455	0.649	(−0.001, 0.001)
EQM	−0.000011 **	0.000004	−2.690	0.007	(−0.000019, −0.000003)
PM × ATO	1.0136 ***	0.005	206.76	0.000	(1.004, 1.023)
CORR	−0.0088	0.009	−0.960	0.337	(−0.027, 0.009)
RD	0.0954	0.234	0.407	0.684	(−0.364, 0.555)
PTE	0.0003	0.058	0.005	0.996	(−0.114, 0.115)
UNP	0.0036	0.012	0.314	0.754	(−0.019, 0.026)

Note: * p < 0.1; ** p < 0.05; *** p < 0.01.

includes the coefficients and Significance Test Statistics for Greece during the examined period 2005–2020.

The analysis showed that the variable profit margin (PM) had a positive and statistically significant effect on ROE, with a coefficient of 0.0053 and p < 0.001, suggesting that an increase in profitability is directly related to an increase in ROE. The PM × ATO interaction term is also highly significant, with a large positive coefficient (1.0136) and a very low p-value (p < 0.001), revealing a strong synergistic mechanism between profitability and efficient asset utilization. In contrast, the ATO and EQM variables showed opposite effects: ATO is not statistically significant (p = 0.649), while EQM showed a small but statistically significant negative effect (p = 0.007), suggesting that increased leverage may burden the return on equity. The remaining variables (CORR, RD, PTE, UNP) did not show a statistically significant effect (p > 0.10), indicating that they are indistinguishable from zero in terms of their contribution to ROE. Finally, the model constant (−0.0010) was not statistically significant (p = 0.776), with the confidence interval including zero, suggesting that the expected value of ROE in the absence of all variables is negligible.

According to our diagnostic tests, the Durbin–Watson index had a value of 1.783, which is relatively close to two, indicating that there was no strong first-order autocorrelation problem in the residuals. However, because the value was not exactly two and the data were time series, further testing was required to fully confirm the absence of autocorrelation. The results of the Breusch–Pagan test indicated the existence of heteroskedasticity, which necessitated the use of robust weighted standard errors HC0 to draw more reliable conclusions. Furthermore, according to the Shapiro–Wilk and Jarque–Bera tests, the residuals did not follow the normal distribution. For this reason, confidence intervals via resampling (bootstrapped CI) were used, a method that does not rely on the normality assumption and is considered more realistic when the classical OLS assumptions are violated.

Moreover, strengthening our tests in OLS estimation, Variance Inflation Factors (VIF) assessed the level of multicollinearity between the independent variables in the regression model.

These VIF (Variance Inflation Factor) values offer a handy guide for spotting multicollinearity, which is when our independent variables are a bit too “comfortable” with each other. A VIF of one is the ideal, showing no correlation at all. If VIFs are between one and five, there is some moderate overlap, but it is likely not a big deal. However, once the five to 10 range is crept into, that high correlation might start messing with our coefficient estimates. And if a VIF of 10 or higher is hit, a serious multicollinearity situation shows up that could make our coefficient estimates shaky and untrustworthy.

By analyzing these VIF results (

Table 7. OLS with respect to Greece, Variance Inflation Factors (VIFs)—multilinearity—no. of obs.: 2.555.

Variable	VIF	Interpretation
const	131.05	Very high (perhaps due to scale/intercept)
PM	1.27	Low (ok)
ATO	1.04	Low (ok)
EQM	1.03	Low (ok)
PM × ATO	1.41	Low (ok)
CORR	2.22	Moderate
RD	4.98	High (attention)
PTE	13.15	Very high (multilinearity)
UNP	7.24	High (attention)

), the constant term came in with a hefty VIF of 131.05, but as noted, this is often the case and usually not a red flag for interpreting other coefficients. Profit margin (PM) at 1.27, asset turnover (ATO) at 1.04 and the equity-to-equity ratio (EQM) at 1.03 all showed impressively low correlation with other variables. Their interaction term, PM × ATO, also sits at a comfortable 1.41. Corruption (CORR) has a moderate VIF of 2.22, likely not a major worry. Research and Development (RD) nudged towards the higher end at 4.98, suggesting a strong link with another variable. However, part-time employment (PTE) shoots up to a problematic 13.15, signaling serious multicollinearity. Lastly, unemployment (UNP) at 7.24 also indicates high correlation, potentially causing issues. In short, while most variables seem well-behaved, PTE and UNP are the main culprits raising multicollinearity concerns.

The high Variance Inflation Factors (VIFs) observed for the labor market variables reveal structural multicollinearity. Consequently, the standard errors for these specific coefficients may be overstated, reducing the accuracy of their point estimates and making it difficult to isolate their independent effects on ROE. While this limits the explanation of individual coefficients, the overall model fit proposes that the combined macro-institutional environment has a weaker descriptive power compared to accounting variables. Moreover, this does not abrogate the overall model fit or the robust significance of the fundamental accounting variables.

Time fixed effects were mislaid to avoid over-specification given the inclusion of time-varying macroeconomic variables.

To address heteroskedasticity, a corrected version of the standard errors is used that addresses heteroskedasticity (

Table 8. OLS with respect to Greece, coefficients with durable (HC0) standard errors—no. of obs.: 2.555.

Variable	Robust Std. Error (HC0)
const	0.0032
PM	0.0017
ATO	0.0003
EQM	0.00001
PM × TO	0.0097
CORR	0.0081
RD	0.2280
PTE	0.0590
UNP	0.0109

), in the form of HC0 (heteroskedasticity-consistent) standard errors; when the classical error assumptions are violated (as was done here, according to the Breusch–Pagan test), these corrections are used for more reliable p-values and confidence intervals.

The use of robust standard errors (HC0) enhances the statistical validity of the results, especially in a heteroskedasticity setting. The present analysis shows that only PM and PM × TO variables are reliable and significant predictors of ROE.

The confidence intervals give a range of values within which it is expected that the true value of the coefficient for each variable lies, with 95% probability. The bootstrapping method is an iterative resampling process from the original data, which allows us to estimate the distribution of the coefficients without relying on traditional assumptions (such as normality of the residuals).

The PM and PM × ATO variables were clearly significant, since the intervals did not include zero. The remaining variables either had CIs that include 0 or were too wide, indicating low statistical or practical significance. There are variables with wide CIs (e.g., RD, PTE), indicating that their effect is unstable, perhaps due to multilinearity or heteroskedasticity.

4.2.2. Significance of Variables

The analysis identified the PM × ATO variable (

Table 9. OLS with respect to Greece, bootstrapped 95% confidence interval—no. of obs.: 2.555.

Variable	95% Bootstrapped CI
const	(−0.0073, 0.0058)
PM	(0.0024, 0.0090)
ATO	(−0.0008, 0.0005)
EQM	(≈0, ≈0)
PM × ATO	(0.9930, 1.0325)
CORR	(−0.0256, 0.0074)
RD	(−0.3526, 0.5110)
PTE	(−0.1134, 0.1162)
UNP	(−0.0172, 0.0263)

) (profitability and asset return product, according to DuPont analysis) as the dominant determinant of ROE, with a coefficient of 1.0136 (p < 0.001), maintaining high statistical significance and a narrow confidence interval even after bootstrapping (95% CI: 0.9930–1.0325).

The profit margin (PM) variable also emerged as statistically significant (p = 0.002), with a positive effect on ROE. In contrast, the variables ATO, EQM, CORR, RD, PTE and UNP showed no statistically significant effect, either with standard error values or with robust estimates. The overall picture of the model confirmed its high explanatory power, but only two variables seemed to contribute significantly to the explanation of the return on equity. PM × ATO, which composes the key parts of the DuPont model, confirmed its theoretical role as the main determinant of ROE.

Continuing with the firms in Cyprus, OLS regression results are presented below. Again, two sets of results were presented, one with “nonrobust” standard errors and the other with “HC0” (White’s heteroskedasticity-consistent) standard errors. Both of them were analyzed to see the differences.

4.2.3. Cyprus

Analyzing the Cyprus companies, based on the OLS results (

Table 10. Coefficients and statistical significance, Cyprus, 2005–2020—no. of obs.: 474.

Variable	Coefficient (Nonrobust)	Standard Error (Nonrobust)	p-Value (Nonrobust)	Coefficient (HC0)	Standard Error (HC0)	p-Value (HC0)
const	−0.0219	0.023	0.343	−0.0219	0.027	0.415
PM	0.0096	0.002	0.00	0.0096	0.007	0.151
ATO	0.0017 *	0.002	0.289	0.0017	0.001	0.055
EQM	~0	1.7 × 10−5	0.567	~0	3.22 × 10−5	0.762
PM × ATO	0.9731 ***	0.019	0.00	0.9731	0.042	0.00
CORR	0.0204	0.036	0.570	0.0204	0.040	0.614
RD	0.8602	1.507	0.568	0.8602	2.210	0.697
PTE	−0.0892	0.127	0.482	−0.0892	0.140	0.523
UNP	0.0801	0.085	0.347	0.0801	0.082	0.330

Note: * p < 0.1; ** p < 0.05; *** p < 0.01.

), the following results are obtained. The observations consisted of 474 entries and concern 74 firms for the period under consideration, 2005–2020.

According to OLS results for Cyprus presented, the R-squared is 0.911. This means that the model explained 91.1% of the variance in ROE. In other words, the independent variables together explained very well the changes in ROE of Cypriot companies for the period 2005–2020. The adjusted R-squared was very close to the R-squared, suggesting that the addition of independent variables to the model is justified and not simply due to an increase in the number of variables.

The analysis included analysis of coefficients (Coef), standard errors (Std Err), t-statistic (t) or z-statistic (z), p-value (p > |t| or p > |z| and confidence intervals ([0.025 0.975]).

The analysis showed remarkable differences between the two models (with and without robust errors). The constant (const) was not statistically significant in either model (p > 0.05), indicating that it does not significantly affect ROE when all other variables are zero. Profit margin (PM) showed a statistically significant positive effect in the model with simple errors (p = 0.000); however, this significance disappeared in the model with robust HC0 errors (p = 0.151), suggesting that the original estimate may have been affected by a violation of OLS assumptions—an important finding. ATO was positive but not statistically significant in both models, although in the HC0 model the p-value (0.055) was marginally close to 0.05, indicating a possible effect that deserves further investigation. EQM had a very small and non-significant coefficient, while the variables CORR, RD, PTE and UNP remained insignificant in both models. Finally, the PM × ATO interaction was positive and highly statistically significant in both models (p = 0.000), supporting the idea that the combination of profitability and asset efficiency has a significant and consistent effect on ROE, regardless of the use of simple or robust estimators.

While the high R-squared values raise possible overfitting concerns for both countries, the coherence of the results across different specifications (OLS, robust errors, GMM) and the use of bootstrapping suggest that the core relationship (PM × ATO) is robust.

▪

Autocorrelation: The Durbin–Watson statistic shows at 1.626 (

Table 11. OLS with respect to Cyprus, diagnostic tests for the model.

Statistical	Value	Interpretation
Durbin–Watson	1.626	Slightly autocorrelated residuals (close to two = okay).
Breusch–Pagan (Heteroskedasticity)	p = 8.52 × 10−6	There is heteroskedasticity, so we used HC0 in the second model.
Shapiro–Wilk/Jarque–Bera	p < 0.0001	Residuals do not follow a normal distribution.

). Generally, values around two suggest that the residuals (the differences between the observed and predicted values) are not significantly correlated with each other. While 1.626 is not far away from two, it does lean towards positive autocorrelation (where consecutive residuals tend to have the same sign). The commentary rightly advises caution, especially when dealing with time series data.

The Shapiro–Wilk and Jarque–Bera tests (p-value < 0.0001) strongly rejected the null hypothesis of normality for the residuals. Consequently, to ensure valid statistical inference, robust (HC0) standard errors and bootstrapped confidence intervals were employed.

By analyzing the VIFs (

Table 12. OLS with respect to Cyprus, Variance Inflation Factors (VIFs)—multilinearity.

Variable	VIF	Interpretation
const	335.17	Very high (perhaps due to scale/intercept)
PM	1.53	Low (ok)
ATO	1.05	Low (ok)
EQM	1.04	Low (ok)
PM × ATO	1.59	Low (ok)
CORR	1.20	Low (ok)
RD	2.18	Moderate
PTE	9.61	Very high (multilinearity)
UNP	7.98	High (attention)

) for Cyprus’s firms the next results emerge. The constant term had a very high VIF of 335.17; while seemingly alarming, such a result is often a byproduct of how models are structured, especially when interaction terms are involved and variables might be centered. It generally did not throw shade on the reliability of the other variables’ coefficients. Moving to the individual predictors, profit margin (PM) at 1.53 and the interaction term PM × TO at 1.59 showed moderate, likely unproblematic, correlations. Asset turnover (ATO) at a mere 1.05, equity multiplier (EQM) at 1.04 and corruption (CORR) at 1.20 all boast very low VIFs, indicating minimal overlap with other predictors. Research and development (RD) at 2.18 also fell into a comfortable range of moderate correlation.

On the other hand, unemployment (UNP) at 7.98 and part-time employment (PTE) with a VIF of 9.61 raised concerns. Significant multicollinearity was indicated by these high VIF values, indicating that PTE and UNP had a close relationship with one or more of the other independent variables in the model. Because of this interdependence, it could be challenging to distinguish between PTE and UNP’s distinct effects on ROE. It could cause these variables’ coefficient estimates to become unstable and their standard errors to increase, which could make it more difficult to assess their actual statistical significance. It could be necessary to combine the highly correlated variables and employ other sophisticated regression procedures to address the multicollinearity.

Following our analysis, in accordance with confidence intervals,

Table 13. OLS with respect to Cyprus, bootstrapped 95% confidence interval.

Variable	95% Bootstrapped CI
const	(−0.0756, 0.0273)
PM	(−0.0016, 0.0231)
ATO	(−0.0001, 0.0033)
EQM	(−0.0001, 0.0001)
PM × TO	(0.8915, 1.0474)
CORR	(−0.0603, 0.0966)
RD	(−3.9704, 4.7559)
PTE	(−0.3508, 0.2150)
UNP	(−0.0958, 0.2491)

, it can be said that the bootstrapped confidence intervals were generally consistent with the conclusions drawn from the analysis of p-values in the model with robust standard errors (HC0). The PM × ATO interaction is the only variable that consistently showed a statistically significant effect on ROE, as its confidence interval did not include zero.

The other independent variables (including PM, ATO, EQM, CORR, RD, PTE, UNP) did not appear to have a statistically significant direct effect on ROE, as their confidence intervals include zero. The borderline case of ATO needs attention.

5. Discussion

This research underscored that the profitability of firms in Greece and Cyprus is based primarily on internal economic metrics—primarily profit margins and asset utilization efficiency. Our study using DuPont and regression techniques corroborates that these two variables, particularly their interaction (PM × ATO), were powerful and reliable predictors of ROE. On the other hand, external non-financial factors such as corruption, unemployment and even R&D intensity did not show statistically significant effects.

This does not imply that non-financial variables are irrelevant, but that they may be indirect, protracted, or insufficiently reflected in the current data. For example, an investment in R&D may require a longer period to be reflected in profitability. Similarly, governance and labor practices may influence other performance metrics before they affect ROE.

No additional robustness analysis was performed, as the reliability of the results has already been ensured through the application of the System GMM method and the relevant tests (Arellano–Bond test for autocorrelation and Hansen tests for instrument validity). Furthermore, the estimates are based on robust standard errors, which minimize the effects of heteroskedasticity and multicollinearity.

The financial sector was not excluded from the sample, as it is an essential part of economic activity in the countries under review (Greece and Cyprus) and contributes significantly to overall corporate profitability and the transmission of macroeconomic outcomes (Athanasoglou et al., 2008; Dietrich & Wanzenried, 2011). Its exclusion would lead to a partial loss of information and possible sample selection bias.

Given these parameters, it was not deemed necessary to perform a separate robust analysis with or without the financial sector, as its inclusion has already been tested econometrically and is considered reasonable based on the research objective.

6. Conclusions

The results of this study showed the comparative relevance and dominance of accounting versus non-financial indicators. In particular, profit margin (PM) and asset turnover (ATO) were the most important determinants of corporate profitability, confirming the traditional financial view of the DuPont model (Nissim & Penman, 2001; Soliman, 2008; Barbier, 2020; Wahlen et al., 2023). Similar results have been reported in recent studies on developed and emerging markets, which indicate that internal operational variables explain profitability to a greater extent than external macroeconomic factors (Hristov et al., 2023).

In contrast, non-financial variables such as corruption, unemployment and part-time employment do not appear to be statistically significant. This finding is consistent with the observations of Hoinaru et al. (2020), who showed that, in European countries, corruption indirectly affects corporate performance through institutional quality rather than directly through accounting profitability. Similarly, Mazzi et al. (2019) argue that institutional indicators often act as long-term factors, which may explain the lack of short-term statistical significance in the present sample.

Furthermore, the results were consistent with the predictions of the Resource-Based View (RBV), according to which the internal resources and capabilities of a company (such as operational efficiency and effective capital management) are decisive for the creation of value and sustainable profitability (Barney, 1991; Peteraf, 1993). At the same time, the relative weakness of institutional variables was consistent with Institutional Theory (North, 1990), which predicts that institutional characteristics become more important in periods of institutional reform or instability—situations that are limited in time in the study sample.

Unlike prior studies in developed markets that often find a direct link between macro-institutional stability and firm performance (Eldomiaty et al., 2023; Hoinaru et al., 2020), this study’s results indicate that in crisis-affected open economies like Greece and Cyprus, internal operational efficiency (PM × ATO) acts as the primary neutralizer, decoupling firm profitability from immediate external shocks, aligning with the core tenets of the Resource-Based View (Barney, 1991; Soliman, 2008).

In comparison to studies on Southern European countries, this study showed that companies in Greece and Cyprus are more dependent on internal accounting factors and less influenced by institutional variables. Thus, this study failed to find a statistically significant direct relationship in the short term. This may reflect the relative homogeneity of institutional conditions and the lower differentiation of labor markets in the two economies.

According to the study’s findings, it is critical for businesses to invest in Strategic Focus and Long-Term Vision. It is suggested that companies incorporate non-financial variables into their strategy with a long-term view. For example, investment in Research and Development (RD) should be considered a strategic choice to create competitive advantage and future profitability, even if it does not have an immediate return.

Also, it would be better for firms to focus on Creating and Maintaining a Strong Corporate Reputation. A positive reputation, corporate social responsibility and ethical business behavior can enhance the trust of stakeholders (customers, investors, employees), leading to increased loyalty, better access to talent and capital and ultimately higher profitability. Companies can invest in CSR initiatives, transparency and communication of their values.

Furthermore, investing in Training and Staff Development is another way for firms to enhance the impact of non-financial variables on profitability. A well-trained and skilled workforce is vital for innovation, efficiency and product/service quality. Companies need to continuously invest in developing the skills of their employees by creating a continuous learning environment. This can lead to improved productivity and competitiveness, positively impacting profitability in the long term.

In addition, it is vital for firms to base their business on Uncertainty and Unpredictable Factors, so as to reinforce the influence of non-financial variables on profitability. The ability of a company to deal with unpredictable risks and adapt to a changing environment is critical to maintaining profitability. Companies can invest in risk management systems, flexibility in their operations and developing supply chain resilience.

In a further step, improving Measurement and Monitoring by firms may strengthen the contribution of non-financial variables to profitability. Companies need to improve how they measure and monitor non-financial variables. The use of qualitative and quantitative indicators linked to specific business objectives will allow for a better understanding of their impact on performance.

Moreover, this study recommends firms concentrate on Integration into Integrated Models. In the future, companies could seek to incorporate non-financial variables into more integrated models for forecasting and managing profitability, considering potential time lags and indirect effects.

Although non-financial variables (CORR, RD, PTE, UNP) did not have a direct, statistically significant effect on profitability during the period considered for both countries, this does not mean that they are insignificant for the future profitability of Cypriot companies. The non-significance may be due to several reasons, such as time lag and indirect relationships. The association of these variables with profitability may not be immediate but may occur with a time lag. For example, investments in Research and Development (RD) may not yield immediate profits but may lead to innovative products and services that will increase profitability in the long run. Also, indirect effects play an important role. Non-financial variables may affect profitability indirectly through other factors. For example, a strong corporate reputation (possibly related to CORR) could lead to increased customer and investor confidence, which consequently positively affects sales and access to capital.

It is crucial for Cypriot and Greek firms to recognize the strategic importance of non-financial variables for long-term sustainability and profitability, even if such variables had no immediate statistically important effect on profitability during the period under review. Conscious investment and management of these factors can create competitive advantages and lead to better financial performance in the future.

From a practical standpoint, firms in both countries should prioritize operational efficiency and strong cost management strategies. At the same time, regulators and policymakers should aim to improve institutional quality and labor market structures, which might enhance firm profitability over the long run.

Limitations and Future Research

This search includes 263 firms across Greece and Cyprus, which might be a signal that not all of their economies’ companies (Greece and Cyprus) are captured. Also, non-financial variables, just like the ones included in this study, may have long-term impacts, which are not fully examined in this timeframe.

A noteworthy restriction of this study is the dependence on aggregate country-level proxies (e.g., national R&D expenditure, corruption index) due to the unattainability of consistent firm-level non-financial data for the entire sample period (2005–2020) in Greece and Cyprus. This may not reflect individual investments or strategies at the firm level. While firm-specific metrics would be desirable, the aggregate variables attend as proxies for the broader institutional and macroeconomic ‘pressure’ applied on all firms.

Future research could use firms’ data for non-financial factors instead of each country’s, as has been done in this study. In addition, the post-COVID period is also to be studied. Future research could extend the analysis to 2025 to examine the impact of the COVID-19 pandemic and recovery policies on corporate performance, using data from 2021 onwards to search how these factors influenced each country’s or firm’s profitability. Additionally, future researchers can take the approach of cross-country comparisons. A comparison between Greece/Cyprus and other EU members or economies in the Mediterranean can take place, in order to test if these observed patterns are regionally singular or more global.

The reasonably low influence of the tested factors indicates a clear path for future research, even though this study purposefully concentrated on structural and institutional variables such as unemployment and corruption to analyze the larger theoretical framework of Institutional Theory. Future researchers may extend this approach by including macroeconomic and non-financial elements like inflation rates and currency rate volatility that are more directly related to a firm’s daily operation. A more detailed understanding of the current external limitations affecting firms’ profitability may be obtained by looking at these operationally important elements in addition to core accounting drives.