Profitability and ESG performance in tourism

Sustainable tourism has become a central concern for firms, regulators and investors, and is increasingly operationalized through environmental, social and governance (ESG) metrics. At the same time, tourism firms operate in a capital-intensive and cyclical environment in which shareholder profitability is shaped not only by operating performance but also by financing choices and exposure to demand shocks that can amplify gains or accelerate distress. In this setting, a clear understanding of how accounting fundamentals map into return on equity (ROE) provides a necessary reference point for interpreting whether and under what conditions ESG assessments are associated with financial outcomes, and whether they are reflected in market valuation.

To disentangle accounting-driven profitability dynamics from ESG-related associations, this study follows a two-step design. First, it characterizes shareholder profitability in tourism firms through a DuPont-based decomposition, estimating a fixed-effects ROE model to assess how within-firm changes in accounting components relate to shareholder profitability. Second, building on this framework, the study examines ESG performance for the subsample of firms with available Refinitiv data, testing whether ESG performance is associated with subsequent shareholder profitability (ROE), operating profitability and market valuation.

Empirically, the DuPont-based analysis uses accounting data for 1,618 publicly listed tourism firms over the 2020–2024 period. ROE is modelled as a function of ROA, leverage (total assets divided by equity) and an interest-and-tax-burden ratio measured as net income (NI) divided by earnings before interest and taxes (EBIT). A sign-aware double-negative indicator is also included to flag firm–years in which both NI and EBIT are negative. All continuous accounting variables are winsorized at the 1st and 99th percentiles. The model is estimated with firm and year fixed effects, and standard errors are clustered at the firm level.

ESG analyses focus on 363 distinct firms with ESG coverage and employ three specifications. In all ESG specifications, ESG performance is measured in the prior year. A first ESG analysis examines whether prior-year ESG performance explains within-firm variation in ROE, controlling for accounting variables and firm and year fixed effects. The study then estimates pooled OLS models with operating profitability (ROA) as the dependent variable to assess whether higher prior-year ESG performance is associated with operating profitability in the pooled cross-section. The aggregate ESG score is also replaced with the three pillar scores, each measured in the prior year, to identify which dimension is most closely related to operating profitability. Finally, a pooled OLS valuation model relates the logarithm of market capitalization to prior-year ESG performance, controlling for firm size and year fixed effects.

In this way, the article addresses the following research questions. What drives ROE in listed tourism firms, in terms of its underlying accounting components? Does prior-year ESG performance predict subsequent ROE? Does ESG performance link to operating profitability one year ahead, and which ESG pillar matters most? Is prior-year ESG performance associated with higher market capitalization?

The findings can be summarized as follows. The ROE analysis shows that shareholder profitability in listed tourism firms is tightly linked to operating performance and severe-loss regimes, whereas leverage may have a negative association. Turning to ESG analyses, the results show no evidence that higher ESG performance in one year is followed by higher shareholder profitability for the same firm in the following year. By contrast, stronger ESG performance is associated with higher operating profitability in the subsequent year, mainly through the environmental and social pillars rather than governance. Moreover, ESG performance measured in the prior year is positively related to market capitalization, suggesting that sustainability is reflected in how tourism firms are valued by investors.

Accordingly, the study contributes to the tourism literature by separating a DuPont-based ROE model from the ESG analyses, allowing within-firm dynamics to be distinguished from cross-sectional associations and by extending the evidence to market valuation through market capitalization. In line with the two-step empirical design, the article is structured as follows. Section 2 reviews the related literature and positions the article’s contribution. Section 3 describes the data, sample construction, variables and empirical strategy for the ROE analysis. Section 4 presents the ESG subsample, econometric specifications and results. Section 5 concludes and highlights implications and limitations.

In recent years, a growing body of research has examined the relationship between ESG and financial performance, with results that remain mixed and highly context-dependent. Using a value-at-risk framework, Capelli et al. (2023) show that incorporating ESG risk information improves the prediction of extreme portfolio losses, especially during periods of market stress. Evidence from BRICS markets suggests a positive association between overall ESG scores and financial performance, while the individual environmental (E), social (S) and governance (G) pillars are not consistently associated, implying that composite measures may capture joint effects not visible in disaggregated indicators (Yilmaz, 2021). Other studies focus on ESG risk and value creation, finding that higher ESG risk is associated with higher compliance costs and lower firm value, while disclosure practices such as materiality reporting may mitigate these adverse associations (Eriandani and Winarno, 2024; Puente De La Vega Caceres, 2024). Complementary studies highlight the role of measurement and governance processes, showing that analysts’ assessments often emphasize governance quality and disclosure, and that firm-level governance structures and leadership traits are linked to sustainability performance (Mandas et al., 2023; Naciti, 2019; Venugopal et al., 2023).

Similarly, a growing tourism-sector literature examines the relationship between ESG performance and financial performance, investigating whether ESG metrics capture underlying profitability fundamentals or instead entail trade-offs with them. Evidence from international samples suggests that stronger ESG performance or more extensive ESG disclosure relates to lower firm risk and, in some settings, to a lower cost of equity, consistent with stakeholder-related and risk-reduction mechanisms that may support profitability over time (Abdelsalam et al., 2025; Kumar, 2024; Salvi et al., 2024).

The literature underscores that ESG matters in tourism, but not in a simple way. According to the analysis conducted by Uyar et al. (2020) in the hospitality and tourism industry, firms with a corporate social responsibility (CSR) committee and female directors exhibit higher CSR performance across the E, S and G pillars. However, in their tests, higher CSR performance does not translate into significantly better financial performance. In a closely related tourism-services setting using Refinitiv scores, Matsali et al. (2025) provide post-crisis panel evidence that the E, S and G pillar scores are each significantly negatively associated with ROA, while the ESG controversies score is not statistically significant, suggesting that ESG engagement may entail short-term profitability costs in tourism services. Additionally, Bodhanwala and Bodhanwala (2022) argue that ESG’s performance implications are both pillar-specific and subsector-specific, since hotels, transportation and leisure exhibit heterogeneous responses, suggesting the value of targeted ESG strategies.

Longitudinal evidence shows that hospitality firms invest more in environmental initiatives and face fewer environmental concerns than comparable industries, while stronger financial performance enables sustainability investment that is subsequently associated with improved performance, consistent with a virtuous cycle (Singal, 2014). Buallay et al. (2022) find that higher ESG engagement and transparency improve operational efficiency and market valuation, but the relationship is non-linear, with diminishing returns beyond a certain level and no clear effect on shareholder profitability. Furthermore, Su and Chen (2020) show that hospitality firms’ ESG commitments are associated with positive stock market reactions. In addition, using a sample of 274 listed Chinese tourism firms, Zhang et al. (2025) find that employment levels are a strong predictor of ESG performance, whereas lagged Tobin’s Q is negatively associated with ESG, indicating a potential trade-off between market valuation and ESG engagement. This pattern of mixed and potentially non-linear payoffs is consistent with the corporate sustainability literature, which highlights firms’ trade-offs between capturing private value and generating broader shared value, alongside persistent tensions between sustainability measurement and effective managerial implementation, with financial returns that may therefore emerge only gradually or remain difficult to observe (Haffar and Searcy, 2017).

According to Kumar (2023), tourism firms use sustainability disclosure to build stakeholder relations and to dampen the effect of macro-level uncertainty on firm value during crisis periods. Based on an analysis of ESG reports from the 50 largest hospitality and tourism firms, Kim et al. (2025) document substantial variation in disclosure across industries, regions, and business scopes, and highlight recurring concerns about definitions, measurement validity and reliability that undermine comparability. Similarly, Bernard et al. (2025) review ESG reports in hospitality and find that many firms communicate sustainability in an external, fragmented way that overlooks the sector’s diverse stakeholders and operational realities. Beyond disclosure comparability, what ESG captures is also a methodological issue.

Accordingly, in the broader sustainability debate, two issues remain central and contested: the link between financial performance and ESG performance, and the substantial divergence across ESG ratings. On the measurement side, there is substantial rating divergence among six prominent ESG rating agencies: Kinder, Lydenberg and Domini (KLD); Sustainalytics; Moody’s ESG (Vigeo-Eiris); S&P Global (RobecoSAM); Refinitiv (Asset4) and MSCI (Berg et al., 2022). Other comparative work reports high correlations among KLD, Sustainalytics and Asset4, with broader variable and international coverage for Sustainalytics and Refinitiv (Harrison et al., 2023). By contrast, other evidence shows relatively high correlations between ratings provided by Asset4 and Sustainalytics, while MSCI’s ratings exhibit negative correlations with both (Bissoondoyal-Bheenick et al., 2024). A stronger critique emphasizes the very low reliability and inter-agency agreement among leading ESG rating providers for S&P 500 firms, raising concerns about the usefulness of ESG ratings for investment decision-making (Charlin et al., 2024). For instance, Zhao and Murrell (2022) find that corporate financial performance (CFP, measured by ROE) may causally precede improvements in corporate social performance (CSP) when CSP is measured with KLD ratings, whereas evidence for a causal effect of CSP on CFP is weak. However, when CSP is measured using Sustainalytics scores, they find no causal relationship in either direction.

Although the association between ESG factors and financial performance has captured noteworthy attention from researchers in recent years, financial performance seems to remain firms’ primary concern (Nafi, 2025). ROA and ROE are among the most widely used measures of financial performance (Griffin and Mahon, 1997), and the DuPont analysis is one of the most commonly employed frameworks for decomposing and assessing that performance (e.g. Doorasamy, 2016). The DuPont framework decomposes ROE into key components that link the income statement and balance sheet, illustrating how operating profitability and asset efficiency translate into returns to equity. In the classic three-factor DuPont decomposition, ROE can be expressed as the product of the net profit margin, asset turnover and financial leverage (Moro-Visconti, 2024).

Several scholars have emphasized the importance of the DuPont model for forecasting future profitability and for providing predictive information (Nissim and Penman, 2001; Soliman, 2008). Recent works have used the DuPont framework to decompose and compare the most important profitability drivers. Analysing a large sample of public companies through the DuPont framework, Jape and Malhotra (2023) found that return on capital employed (ROCE) is the most important driver of ROE. Kumar et al. (2024) showed that, among the DuPont levers, efficient working capital management most strongly shapes how profit margin translates into ROE. Jin (2017) used the DuPont analysis to examine changes in ROE following mandatory IFRS adoption in Canada. Further studies have employed the DuPont model to explain ROE and its components in emerging markets (Aryantini and Jumono, 2021; Endri et al., 2020) and among SMEs (Shabani et al., 2021).

More recently, Jin et al. (2022) applied the DuPont framework to quantify the performance impact of COVID-19 on Chinese high-tech firms. Many other studies have employed the DuPont framework for industry analysis by exploring profitability and its determinants in healthcare and hospitals (Bai et al., 2022; Turner et al., 2015), in the energy sector (Baran et al., 2022; Bunea et al., 2019; Kusz et al., 2022), in the luxury industry (Bothra and Gupta, 2020) and among manufacturing firms (Weidman et al., 2019).

Conversely, the use of the DuPont framework in tourism is relatively rare, with a few notable exceptions. For example, Poretti and Heo (2022) apply a DuPont lens to listed tourism firms and show that, before COVID-19, firm value was driven primarily by asset efficiency, with both profit margin and asset turnover predicting value one year ahead. In addition, Akbulaev et al. (2020) combine Altman and DuPont approaches to analyse the bankruptcy of a well-known travel company, the Thomas Cook Group, employing ROE decomposition to diagnose the drivers of insolvency. Moreover, with rare exceptions, few studies have used the DuPont analysis to examine the relationship between ESG performance and financial indicators (Hu et al., 2024; Lei et al., 2025).

This study seeks to fill this gap by applying the DuPont framework to characterize how operating profitability and capital structure map into shareholder returns in the tourism sector, and then testing whether Refinitiv’s ESG score is associated with these accounting fundamentals within a firm fixed-effects framework. Refinitiv’s ESG score is a composite indicator summarizing firms’ relative performance on environmental, social and governance dimensions, based on standardized and publicly disclosed information. The score ranges from 0 to 100, with higher values denoting stronger ESG performance, and is interpreted as an overall sustainability assessment. Following Matsali et al. (2025), Refinitiv’s ESG scores are interpreted as informative indicators of firms’ sustainability strategies and related disclosure practices. The present study builds on the premise that jointly considering financial and ESG performance is increasingly regarded as important for effective managerial decision-making.

The DuPont framework is used to characterize shareholder profitability in tourism firms and to structure the subsequent ESG analyses. The objective of the ROE analysis is to assess whether the relationship between operating profitability, capital structure and equity returns is stable in the data, once extensive heterogeneity across firms and contexts is controlled for. Starting from the DuPont formulation (Moro-Visconti, 2024), ROE may be decomposed into the product of ROA, financial leverage and an interest-and-tax-burden ratio, measured as net income (NI) divided by earnings before interest and taxes (EBIT). To translate this framework into measurable constructs, the study combines accounting data with externally provided ESG scores. The variables used in the empirical models are summarized in Table 1. Accounting data are extracted from Orbis (Bureau van Dijk) and reported in USD thousands; all ratios are computed from the extracted accounting variables rather than using Bureau van Dijk’s pre-calculated ratios. ESG scores are obtained from Refinitiv. Descriptive analyses were conducted in IBM SPSS Statistics (version 26), while regression analyses and robustness checks were performed in R using RStudio.

The relationship between ROE and ESG is examined for publicly listed tourism companies. This choice is motivated by the higher likelihood of ESG coverage for such firms. The sample includes firms for which all values for total assets, equity, EBIT and net income (NI) were available over the 2020–2024 period. The resulting sample consists of 1,618 very large firms. The sectoral composition of the firms in the sample is illustrated in Table 2, which reports a cross-tabulation of the sample by NACE class. The sectoral composition of the sample is dominated by hotels and accommodation services and food and beverage activities, followed by travel agencies and reservation services and passenger transportation (air, rail and maritime). Overall, the distribution reflects the core structure of the tourism value chain, with a strong presence of lodging, hospitality and mobility-related activities.

The sample exhibits substantial geographical concentration, as reported in Table 3. Firms are predominantly located in China (186 companies) and Japan (159), followed by the United States (113) and India (105). A non-negligible share of firms is incorporated in the Cayman Islands (88), reflecting the widespread use of offshore financial centres in tourism-related and transportation industries. Southeast and East Asian countries are also well represented, including Vietnam (82), Indonesia (74), Taiwan (52) and Thailand (37), while the United Kingdom (39) is the only European country among the top 10. Overall, the geographical distribution underscores the strong exposure of the sample to Asian markets and emerging economies.

Table 4 summarizes the descriptive statistics for the variables used in the ROE regression, computed on pooled firm-year observations over 2020–2024 for 1,618 firms. After winsorization at the 1st and 99th percentiles, shareholder profitability, measured by ROE, shows high dispersion, which is consistent with substantial heterogeneity in equity returns across tourism firms. Moreover, ROE has a slightly negative mean, despite a positive median. This suggests that the typical firm remains profitable, while a small number of severe underperformers pull down the mean, particularly during the pandemic period. Instead, operating profitability measured by ROA is more stable and less dispersed, suggesting that part of the variability in ROE reflects financial structure and below-the-line components, rather than operating performance alone. Leverage displays marked heterogeneity, including cases with negative equity and very highly levered firms, both of which signal potential financial fragility in parts of the sample. The median NI/EBIT ratio is 0.769, implying that net income absorbs roughly one-quarter of operating profit for the typical firm-year. The extremely wide dispersion and the presence of large negative values indicate that the gap between operating profit and net income is highly unstable, and driven by episodic shocks in financing, taxation or non-operating components. Moreover, a meaningful share of observations in the firm sample corresponds to severe-loss years in which both net income and EBIT are negative, while firm size (log total assets) shows substantial cross-sectional variation.

Table 5 reports the Pearson correlations among the variables in the ROE specification. ROE is positively related to ROA and negatively related to leverage, indicating that stronger operating performance is associated with higher shareholder returns. Higher indebtedness is associated with weaker ROE, which is also lower in firm-years characterized by severe losses, as captured by the double-negative indicator. By contrast, the association between ROE and the NI/EBIT ratio is economically very small, suggesting that cross-sectional differences in the conversion of EBIT into NI are only weakly reflected in shareholder returns. To assess multicollinearity, variance inflation factors (VIF) were computed for the full ROE specification. All variance inflation factors are close to unity. The VIF for ROA (1.663), for leverage (1.015), for the NI/EBIT ratio (1.117), for the double-negative indicator (1.793) and for firm size (1.050) indicate negligible multicollinearity among the regressors and remain well below commonly accepted thresholds of concern. Overall, the correlation structure and VIF diagnostics suggest that multicollinearity is unlikely to materially affect coefficient estimation or inference in the ROE regressions.

Following the classical three-factor DuPont framework, ROE can be expressed as the product of ROA (EBIT divided by total assets), leverage (total assets divided by equity) and the NI/EBIT ratio (Moro-Visconti, 2024). Beyond this accounting identity, the empirical analysis conducted in this study adopts a linear additive specification to enable statistical inference, fixed-effects identification and the inclusion of indicators capturing loss regimes. Specifically, the baseline specification is estimated on a firm-level panel covering the 2020–2024 period, with firms indexed by i and years indexed by t. All continuous accounting variables are winsorized at the 1st and 99th percentiles to mitigate the influence of extreme outliers. To address unobserved, time-invariant firm heterogeneity, the model is estimated using firm fixed effects, with year fixed effects and standard errors clustered at the firm level. This approach controls for persistent factors such as managerial style, organizational culture, long-run business model characteristics and stable aspects of asset structure. The baseline model is specified as follows:

where, $ROAit$ reflects contemporaneous operating profitability, $Leverageit$ measures the extent to which the capital structure amplifies operating performance into equity returns, and the $NIit/EBITit$ ratio captures the translation of operating profits into bottom-line earnings by incorporating financing costs, taxation and non-operating items. Because the $NIit/EBITit$ ratio can be mechanically positive when both NI and EBIT are negative, the model includes an explicit double-negative indicator, $1(NI<0∧EBIT<0)it$⁠, to isolate severe loss regimes and avoid interpreting as positive those observations where the ratio is positive only because both the numerator and denominator are negative. $Sizeit$⁠, measured as the logarithm of total assets, controls for scale-related differences in profitability dynamics. Moreover, in the empirical specification, $αi$ denotes firm fixed effects and $λt$ captures common macroeconomic shocks and other time-specific conditions affecting all firms. The error term $εit$ reflects idiosyncratic shocks that vary within firms over time, with inference based on heteroskedasticity-robust standard errors, clustered at the firm level. While fixed effects do not address bias from time-varying omitted variables or simultaneity, they substantially mitigate confounding arising from persistent cross-sectional differences across firms.

Table 6 reports the estimates from the firm fixed-effects specification in Eq (1) that operationalizes the theoretical ROE decomposition discussed before. By exploiting within-firm variation over time, the model isolates how changes in operating performance, financial structure and severe loss regimes translate into changes in shareholder profitability, net of time-invariant firm characteristics and common macroeconomic shocks.

The results show that ROA is positive and highly significant, indicating that within-firm improvements in operating profitability are strongly associated with higher equity returns. Quantitatively, a one-unit increase in ROA is associated with an increase of approximately 1.24 in ROE. Leverage, instead, is negative and highly significant (β = −0.054, p < 0.001). This finding implies that, within firms, increases in leverage are associated with lower ROE over the sample period. Rather than amplifying operating performance, higher leverage appears to exert a drag on equity returns, plausibly reflecting increased interest burdens, heightened financial risk or adverse financing conditions during the 2020–2024 period. This result is consistent with the notion that leverage can become value-destroying in environments characterized by elevated uncertainty or tightening financial conditions.

Instead, the NI/EBIT ratio is positive but not statistically significant. The lack of statistical significance indicates that below-the-line factors such as financing costs, taxation and non-operating items play a more limited or irregular role in explaining time-series variation in equity returns relative to core operating performance. Conversely, the indicator for double-negative outcomes, capturing firm-years in which both NI and EBIT are negative, enters with a large and highly significant negative coefficient (β = −0.229, p < 0.001). This result highlights the economic importance of severe loss regimes. Even after controlling for ROA and the NI/EBIT ratio, episodes of simultaneous operating and net losses are associated with a substantial reduction in ROE. Firm size is not statistically significant, suggesting that, within firms, changes in scale do not exert a systematic effect on ROE, once operating performance, leverage and severe loss regimes are taken into account.

Overall, the model exhibits strong explanatory power (R-squared = 0.44). A Wald test confirms the joint significance of the regressors, indicating that a substantial fraction of within-firm variation in ROE among tourism firms is explained by ROA, leverage and severe loss regimes. The results provide robust empirical support for the theoretical structure of the ROE model and establish a solid baseline for evaluating the incremental role of ESG performance in subsequent specifications.

Tourism firms operate in a cyclical and capital-intensive environment in which shareholder profitability is shaped by a combination of operating performance, financial structure and exposure to macroeconomic shocks. In this setting, ESG scores can be interpreted as indicators of firms’ strategic orientation, stakeholder engagement and disclosure intensity, which may correlate with business quality and long-run operating efficiency, but do not necessarily translate into short-run improvements in shareholder profitability.

These timing considerations motivate the empirical strategy. ESG performance enters the models using its prior-year values to assess whether sustainability performance in year t−1 is associated with financial outcomes in year t. The analysis begins with a firm fixed-effects ROE model, in which shareholder profitability is related to prior-year ESG performance, controlling for accounting components and year effects, thereby isolating within-firm variation over time. It then turns to operating profitability, relating ROA to ESG performance in the prior year in pooled cross-sectional regressions. In additional specifications, the aggregate ESG score is replaced by the environmental, social and governance pillar scores to assess which dimensions drive the association. Finally, a pooled valuation model links log market capitalization to ESG performance in the prior year, controlling for firm size and year fixed effects.

ESG information is available only for a subset of firms in the accounting panel. ESG data are observed for 363 tourism firms, defined as those reporting ESG scores in at least one year over the 2020–2024 period. Table 7 presents pooled descriptive statistics for the subsample of firms with ESG coverage. Continuous accounting variables are winsorized, while ESG and pillar scores are reported on their original scale. The values of profitability indicate that ROE is slightly negative on average but positive for the median firm-year, implying that a typical firm with ESG scores is profitable, while a subset of loss-making observations pulls down the mean. ROA is instead clearly positive and less volatile, consistent with generally solid operating performance in this group. Leverage remains highly heterogeneous and includes cases with negative equity, pointing to potential financial fragility for a subset of firms. The NI/EBIT ratio shows wide dispersion, indicating that financing costs, taxes and non-operating items can materially affect the conversion of EBIT into NI. The double-negative dummy confirms that a non-negligible share of observations falls into a severe loss regime. Firms with available ESG scores are comparatively large, consistent with ESG disclosure being more common among bigger and more visible companies. This pattern is also reflected in market capitalization, which is high on average but displays substantial cross-sectional variation. Average ESG performance is moderate, with a mean ESG score of 49 on a 0 to 100 scale and a median of 50.4. Looking at the pillars, the average environmental score is 44.9, compared with 51 for social and 50.1 for governance, indicating relatively weaker environmental performance than the social and governance dimensions.

To assess potential selection into ESG score availability, a logit model is estimated with a dependent variable equal to one if a firm has an ESG score. Firm size is included as the main explanatory variable, while country, sector and year fixed effects control for contextual heterogeneity. As shown in Table 8, ESG score availability is strongly and positively associated with firm size (p < 0.001), indicating that ESG scores are not randomly observed across firms but are concentrated among larger firms and systematically vary across countries, subsectors and over time.

As previously mentioned, the ESG model is designed to explore the relationship between shareholder profitability and firms’ sustainability performance by examining whether past ESG performance is associated with subsequent variations in return on equity. By specifying ROE as the dependent variable and measuring ESG performance in the prior year, the framework focuses explicitly on the predictive content of sustainability practices. This approach helps avoid contemporaneous correlations that may reflect reverse causality. All specifications also control for firm and time effects. The use of firm fixed effects ensures that the analysis isolates within-firm changes over time, net of persistent differences across firms such as business models, corporate culture, or industry positioning, while year fixed effects control for common macroeconomic conditions and aggregate shocks. The main empirical specification is given by:

where $ESGi,t−1$ captures ESG performance measured at t−1, while all other controls, fixed effects and standard-error clustering follow the specification described above for Eq. (1).

Moreover, to examine whether firms with stronger ESG performance tend to be more operationally profitable on average, a pooled OLS regression is estimated, using the same one-year lag structure. In this additional regression, ROA is the dependent variable, while the ESG score, measured at t−1, is the key explanatory variable. The model also includes firm size (logarithm of total assets) to control for scale differences across firms and year fixed effects to absorb common macroeconomic shocks and time trends. Year fixed effects are included, with 2021 as the omitted reference year, and heteroskedasticity-robust standard errors (HC1) are reported. The specification is:

where $α$ is a constant, $ROAit$ denotes operating profitability of firm $i$ in year $t$⁠, while $ESGi,t−1$ is the firm’s ESG score measured in the prior year to reduce simultaneity concerns and to allow for delayed effects of ESG policies on operating profitability. $Sizeit$ (log total assets) controls for scale differences across firms, $λt$ captures year-specific factors affecting all firms, and $εit$ is the idiosyncratic error term. In this pooled OLS setting, $β1$ represents the average cross-sectional association between ESG performance and ROA, conditional on size and year effects. All continuous variables (ROA and Size) are winsorized at the 1st and 99th percentiles to mitigate the influence of outliers.

Furthermore, to assess which ESG dimension matters most for operating profitability, the pooled OLS model is re-estimated by replacing the aggregate ESG score with the three ESG pillar scores, each measured one year earlier. As in the baseline specification (Eq. 3), ROA is the dependent variable, firm size is proxied by the natural logarithm of total assets and year fixed effects are included to absorb common macroeconomic shocks. Heteroskedasticity-robust standard errors are used. The three specifications are:

Finally, to examine whether ESG performance is reflected in subsequent firms’ market valuation, the following pooled OLS model is estimated:

where $log(Market_capitalizationit)$ is the natural logarithm of firm $i$’s market capitalization in year $t$⁠, computed after winsorizing market capitalization at the 1st and 99th percentiles. All the other regressors and terms in Eq. (7) are defined as in Eq. (3). Year fixed effects are included, with 2021 as the omitted reference year, and heteroskedasticity-robust standard errors (HC1) are reported.

Table 9 reports firm fixed-effects estimates of the ROE specification in Eq. (2), where the coefficient of primary interest is $β5$⁠, which captures the effect of ESG performance measured in the preceding year. The key result is that the coefficient on $ESGi,t−1$ is statistically insignificant and economically small, once firm and year fixed effects are included, suggesting that within-firm changes in ESG are not systematically followed by changes in ROE over the sample period. Among the controls, ROA is positively associated with ROE, consistent with operating profitability translating into shareholder returns. Moreover, the NI/EBIT ratio is also significant and positively related to ROE, indicating that a higher conversion of EBIT into net income is associated with higher equity returns beyond what ROA captures. Additionally, the double-negative indicator remains negative and significant, confirming that severe-loss regimes drive sharp declines in ROE independently of ESG performance. Conversely, leverage and firm size are not statistically significant in the within-firm specification. The overall R-squared is 0.334, reflecting the substantial share of variation absorbed by firm and year fixed effects.

To address heterogeneity across tourism subsectors, the firm fixed-effects ROE model is re-estimated separately for accommodation, food and beverage, transport, travel agencies and other tourism-related activities. As reported in Table 10, the coefficient on ESG performance measured in the prior year remains statistically insignificant across all tourism subsectors, indicating that the absence of a within-firm ESG–ROE association is not driven by pooling structurally different business models.

The pooled OLS estimates of ROA on ESG performance, as specified in Eq. (3), are reported in Table 11. In this case, the coefficient on ESG performance measured in the prior year is positive and statistically significant, suggesting that firms with higher ESG scores tend to exhibit slightly higher operating profitability in the pooled sample over 2020–2024. Firm size is negatively associated with ROA, and the year indicators for 2022–2024 are positive and significant relative to 2021.

The baseline pooled OLS specification is re-estimated by replacing the aggregate ESG score with the three pillar scores, each measured in the prior year, as shown in Eqs. (4)–(6). As reported in Table 12, the coefficient on the environmental pillar measured in the prior year is positive and statistically significant in explaining ROA, while firm size remains negative and the year effects for 2022–2024 are positive relative to 2021. Table 13 shows a similar pattern for the social pillar measured in the prior year, which is also positive and statistically significant, with the same size and year effects. By contrast, Table 14 indicates that the governance pillar measured in the prior year is economically negligible and statistically insignificant, whereas the control patterns are unchanged. Overall, the pooled OLS results suggest that the positive ESG–ROA association is driven by the environmental and social dimensions rather than governance.

Finally, Table 15 reports pooled OLS estimates of the association between the natural logarithm of market capitalization and ESG performance, as specified in Eq. (7). The coefficient on ESG performance measured in the prior year is positive and statistically significant, indicating that firms with higher ESG scores tend to exhibit higher market valuations after controlling for firm size and year fixed effects. Given the log specification, the estimate implies that, holding firm size and year effects constant, a ten-point increase in the ESG score is associated with approximately a 5.7% higher market capitalization. Year effects are generally not statistically significant, with the exception of a weakly negative coefficient for 2022, likely capturing transient market-wide conditions. The model also shows high explanatory power, with an R-squared of 0.595. Overall, the results suggest that ESG performance is positively priced in the cross section of market valuations.

This study examines how shareholder profitability is structured in listed tourism firms and whether ESG performance is associated with subsequent profitability, once firm heterogeneity is controlled for. The empirical analysis proceeds in two steps. First, a DuPont-oriented panel framework is used to characterize how operating profitability, capital structure, and the NI/EBIT ratio relate to ROE over the 2020 to 2024 period. Second, the analysis turns to ESG performance and uses Refinitiv ESG scores observed in the prior year. It tests whether ESG has predictive content for within-firm changes in ROE in fixed-effects models, and whether ESG is associated with operating profitability in pooled cross-sectional regressions. Finally, the analysis extends to a valuation specification that regresses the log of market capitalization on ESG performance from the previous year.

With respect to the ROE analysis, the fixed-effects results indicate that within-firm variation in shareholder profitability is primarily driven by operating profitability, leverage and severe-loss regimes in which EBIT and net income are simultaneously negative.

As regards the ESG analyses, the results are mixed. In the firm fixed-effects ROE model, the ESG coefficient is economically small and statistically insignificant, indicating no evidence that within-firm improvements in ESG scores are systematically followed by higher shareholder profitability, once time-invariant firm characteristics and year shocks are controlled for. By contrast, pooled OLS estimates with ROA as the dependent variable reveal a positive and statistically significant association between ESG performance and operating profitability, suggesting that firms with stronger ESG profiles tend to be more operationally profitable in the cross-section. When the composite ESG score is replaced by its individual pillars, the environmental and social dimensions remain positively and significantly associated with ROA, whereas the governance pillar does not. Therefore, the profitability link to ESG in tourism firms appears to be driven by environmental and social practices and to reflect primarily cross-sectional differences rather than within-firm dynamics over time. Finally, pooled OLS estimates using log market capitalization indicate a positive and statistically significant association between ESG performance and subsequent market value, after controlling for firm size and year effects. This result suggests that sustainability performance is reflected in cross-sectional differences in investor pricing, implying that ESG performance is valued by financial markets, even when it does not translate into higher short-term shareholder profitability.

In sum, the study’s contributions are to identify the key determinants of profitability in listed tourism firms, starting from a DuPont-based decomposition of shareholder returns, and to assess the relationship between ESG performance and financial performance. In this respect, the results indicate that, once time-invariant firm characteristics are absorbed by fixed effects, ESG does not predict short-run changes in shareholder profitability, whereas pooled comparisons show positive associations with operating profitability and market valuation. This pattern suggests that ESG functions less as a near-term driver of ROE dynamics and more as a marker of persistent capabilities and positioning, as reflected in operating outcomes and investor pricing. The strongest implication is that managers in listed tourism firms should treat ESG primarily as an operational and valuation lever rather than a short-term ROE lever.

As with any empirical study, these findings should be interpreted in light of several limitations. First, the results should be interpreted as associations, not causal effects, because unobserved firm-year shocks may affect both ESG and performance. Second, the ESG analysis relies on a smaller and unbalanced subsample of firms with ESG coverage, which may limit statistical power and generalizability. Third, ESG scores are relatively slow-moving and often embed multi-year information, which may weaken their ability to capture short-run changes in financial outcomes compared with annually measured accounting ratios. Fourth, results may be sensitive to the ESG provider and score construction. Finally, the market-capitalization evidence may reflect investor preferences, ESG-related risk premia or omitted valuation drivers correlated with ESG performance.

To address these issues, future research could extend the analysis using longer ESG panels as they become available, replicate the tests with alternative ESG providers and metrics, and examine longer-run effects and transmission channels, such as cost of capital, risk premia, stakeholder stability, financing constraints or regulatory exposure. More generally, designs exploiting quasi-experimental variation or exogenous shocks would help identify causal mechanisms and assess whether ESG becomes more financially relevant during clearly defined stress episodes or institutional shifts.