The Role of Marketing Intensity in Moderating CSR and Financial Performance in Luxury Fashion ()
1. Introduction
Corporate Social Responsibility (CSR) is a self-regulating business management concept which aims to integrate social and environmental concerns into business operations. In recent years, the implementation of CSR has grown popular in a variety of industries, including the luxury fashion industry. Gucci, for example, has committed to reducing its greenhouse gas emissions by 50% by 2025 as a part of its CSR strategy (Gucci Equilibrium, n.d.). Although prior business management research suggests that CSR detracts from the financial performance (FP) of firms, given that allocating resources towards nonessential “social betterment” policies increases the costs of business operations (Wu & Shen, 2013), other CSR studies outline several theories which rationalize the implementation of CSR.
However, the impact of CSR on the financial performance of designer fashion brands is nuanced due to industry characteristics specific to luxury fashion. Unlike the fashion industry at large, the luxury fashion industry purports values of ostentation, hedonism, and rarity, which are inherently incongruous with CSR goals of sustainability and equity (Kapferer & Michaut, 2015). Despite this subtlety, few studies exploring the CSR-FP relationship in the fashion industry isolate luxury fashion brands. Therefore, to best inform the marketing strategies of designer brands, a study’s exploration of the CSR-FP relationship should account for the nuances of the luxury industry. By investigating the CSR-FP relationship using a sample which entirely consists of luxury fashion brands, the results of this study will significantly inform marketing strategies specific designer brands, allowing luxury fashion companies to benefit stakeholders while simultaneously increasing FP.
Of the few studies which explore CSR in the luxury fashion context, none account for the moderating impact of marketing intensity on financial performance. Prior CSR research across industries suggests that higher marketing expenditure facilitates a link between customers and a company’s CSR initiatives, enhancing consumer awareness of such initiatives. Thus, by incorporating marketing intensity as a moderating variable in analysis, this study enhances the accuracy of the statistical analysis and fills a relevant gap in CSR research.
The findings of this study are limited by the time frame, sample size, the need for the further disaggregation of the CSR dimensions and marketing intensity variable in analysis, and the inability of the researcher to perform advanced statistical analysis.
The remainder of the study is structured as follows: Literature Review; Methodology; Results, Analysis & Conclusions; Limitations & Directions for Future Research; Conclusion; References.
2. Literature Review
2.1. Existing CSR Theories
The three predominant CSR theories in existing literature are the stakeholder, legitimacy, and institutional models (Beqiraj, 2021). The stakeholder theory hypothesizes that companies maximize firm value by creating value not only for shareholders, but all stakeholders involved in business operations. Amongst others, stakeholders in the fashion industry include shareholders, customers, employees, the environment, and governmental and nongovernmental organizations. By implementing CSR, proponents of the stakeholder theory argue that firms create value for stakeholders beyond shareholders, and therefore are able to maximize FP (Thorisdottir & Johannsdottir, 2020).
However, other CSR literature supports the legitimacy model, which proposes that companies implement CSR in order to demonstrate legitimacy to external stakeholders in society. In adhering to societal expectations of business through CSR, some scholars therefore posit that companies are able to cultivate legitimacy and increase brand value (Olateju et al., 2021). On the other hand, the institutional CSR theory suggests that brands implement CSR in response to industry-specific institutional pressure in order to maintain a competitive edge. For example, in the fashion context, the industry-wide push towards a circular economy has led brands to incorporate sustainability practices in business operations (Zeng et al., 2017). While each model proposes a different rationale for the implementation of CSR, the underlying purpose of CSR in each theory is to generate brand value, whether through appealing to stakeholders or responding to institutional pressures. Therefore, in order to examine whether companies benefit from the implementation of CSR, the exploration of the CSR-FP relationship is merited.
Within existing CSR research in the fashion industry, few studies explore the implementation of CSR in the luxury fashion sector in particular. This is significant because the luxury fashion sector is especially subject to public scrutiny due to its nonessential nature (Franco et al., 2019). Thus, research exploring the CSR-FP relationship in this context may elucidate the efficacy of CSR initiatives in mitigating public scrutiny and increasing brand value of luxury fashion companies. Furthermore, in addition to public scrutiny, nongovernmental organizations are increasingly targeting the luxury fashion industry for its unethical and unsustainable practices (Bravo González, 2017). Thus, in line with the institutional theory, luxury fashion brands must develop CSR strategies which meet industry demands in order to optimize FP. Thus, the exploration of the CSR-FP relationship in the luxury fashion context is merited.
2.2. Disaggregated CSR Research in Luxury Fashion
Research exploring the CSR-FP relationship in the fashion industry cannot be extrapolated to luxury fashion due to demographic nuances specific to luxury fashion brands. Unlike the fashion industry at large, luxury brands have historically symbolized hedonism, ostentation, and rarity—the opposite of the equitable and sustainable brand image CSR initiatives aim to cultivate (Kapferer & Michaut, 2015). For example, contrary to the stakeholder theory, professor of CSR and Accountability at the University of Rome Tor Vergata Gloria Fiorani asserts that although the luxury fashion industry is “sensitively exposed to environmental risks,” luxury fashion brands are often the least vigilant in disclosing sustainability practices (Fiorani et al., 2022).
Subsequent research hypothesizes that this is due to the importance of rare materials in establishing the exclusive image of luxury brands. As one study finds, luxury fashion consumers are likely to perceive sustainable products as “cheap” and “lacking prestige” (Vock, 2022). In this instance, the CSR goal of environmental conservation conflicts with the luxury fashion industry’ characteristic exploitation of natural resources. The “sustainable luxury” paradox illustrated in the aforementioned example may thus cause luxury fashion consumers to attribute CSR initiatives to ulterior motives of increasing profit in comparison to other fashion companies (Sipilä et al., 2020). Therefore, the investigation of the CSR-FP relationship in the luxury fashion context is merited independent of existing research on CSR in the fashion industry at large.
Existing research exploring the CSR-FP relationship in the luxury fashion industry is largely inconclusive. For example, as professor of management at the Universita delgi Studi di Torino Laura Broccardo highlights, while (Kapferer & Michaut, 2015) indicates that consumers expect luxury fashion brands to implement CSR, (Dhaliwal et al., 2020) finds that customers do not significantly consider sustainability in purchasing decisions (Broccardo et al., 2022). Although such conflicting results are often attributed to the sustainable luxury paradox, other literature suggests that inconclusive findings are due to the aggregation of the CSR variable in analysis.
Given that CSR is a multidimensional construct, previous findings indicate that the impact of one dimension on FP may counteract that of another (Nollet, Filis, & Mitrokostas, 2016). For instance, although environmental-oriented recycled luxury fashion products may negatively impact FP due to the sustainable luxury paradox, in congruence with the legitimacy theory, social-oriented charitable donations may legitimize the status of luxury fashion brands and thus enhance FP (Sipilä et al., 2020). Furthermore, another study finds that while CSR initiatives targeting external stakeholder may decrease the brand value of luxury firms, CSR policies targeting internal stakeholders, such as employee work-from-home policies, may positively influence FP (Gordano & Chiaudano, 2021). Despite these findings, most studies exploring CSR in the luxury fashion context have either treated CSR as a single, aggregate variable, or solely examined a single CSR dimension in analysis.
Although CSR is most commonly disaggregated into industry-specific dimensions due to variations in CSR implementation across different sectors (Beqiraj, 2021). Therefore, previous inconclusive findings may be the result of the aggregation of the CSR variable or utilization of generic ESG dimensions in analysis. Thus, in order to account for the multidimensional nature of CSR, this study aims to examine the impact of CSR on FP in the luxury fashion context using industry-specific disaggregated CSR dimensions. Although not all CSR dimensions positively impact FP, the significance of the correlation between each CSR dimension and FP remains unexplored as a result of aggregate CSR measures. Therefore, regardless of the direction of the correlation, this study hypothesizes that all disaggregated industry-specific CSR dimensions will have a statistically significant impact on FP in the luxury fashion context.
H1 (Hypothesis 1)—All disaggregated, industry-specific CSR measures will significantly impact the FP of luxury fashion brands.
2.3. Marketing Intensity and CSR
In the luxury fashion industry, customers are the primary target of CSR initiatives. Therefore, in order for CSR to influence consumer behavior and ultimately impact the FP of designer brands, luxury fashion customers must first be aware of the existence of such initiatives. As professor of hospitality strategic management at the Southwest Minnesota State University Yinyoung Rhou explains, “low awareness of CSR initiatives among consumers and other stakeholders undermines CSR effects on firm outcomes” (Rhou et al., 2016). Previous research indicates that higher levels of marketing intensity (MI), the ratio of promotional expenditure to sales, positively influences consumer awareness of CSR, and thus moderates the strength and direction of the CSR-FP relationship. For instance, one study examining CSR amongst U.S. companies found that greater levels of advertisement intensity heightened consumer awareness of CSR initiatives, better disposing customers to purchase the companies’ product (Servaes & Tamayo, 2013; Rahman et al., 2017).
Despite the role of consumer awareness in influencing the efficacy of CSR, there is no prior literature on the role of MI in moderating the CSR-FP relationship in the luxury fashion context. Therefore, this study addresses a relevant gap in research by investigating the research question, “What is the disaggregated impact of CSR on financial performance with the moderating role of marketing intensity?”. If MI is found to moderate the CSR-FP relationship in the luxury fashion context, it would be essential for future studies to incorporate MI as a moderating variable in statistical modeling.
It is necessary to investigate the moderating role of MI in the luxury fashion context irrespective of previous findings due to the sustainable luxury paradox. For example, while (Servaes & Tamayo, 2013) found that greater levels of marketing positively influenced FP amongst a sample of U.S. firms, marketing CSR strategies which conflict with luxury brand attributes may heighten the negative impact of CSR on FP. In this instance, as lower marketing spending can decrease a firm’s CSR visibility (Oh et al., 2014), luxury fashion companies may actually financially benefit from allocating fewer resources towards advertising CSR activities. Regardless of whether luxury firms financially benefit from increased CSR marketing, MI induces greater CSR visibility and thus enhances consumers’ awareness of CSR initiatives. Therefore, this study hypothesizes that CSR will significantly moderate all dimensions of CSR, whether positively or negatively.
H2 (Hypothesis 2)—MI will significantly moderate all dimensions of CSR.
Additionally, given that MI facilitates a link between consumers and CSR initiatives (Rahman et al., 2017), this study posits that including MI as a variable in regression analysis will enhance the accuracy of the statistical models in predicting the CSR-FP correlation.
H3 (Hypothesis 3)—Including MI as a moderating variable will enhance the overall accuracy of the regression analysis in predicting the CSR-FP relationship in the luxury fashion context.
Thus, through employing disaggregated, industry-specific measures of CSR, this study will add to existing literature examining the CSR-FP relationship in the luxury fashion context. This is significant because existing CSR research does not adequately address differences between the values of the luxury fashion industry and the fashion industry at large, and therefore previous findings may not illuminate best marketing strategies for luxury fashion brands to implement. Furthermore, by investigating the role of marketing intensity in moderating the CSR-FP relationship in the luxury fashion industry, this study accounts for the role marketing plays in enhancing consumer awareness of CSR initiatives, a variable which has been omitted from previous CSR research in the luxury fashion industry. This modification will enhance the accuracy of the statistical analysis in predicting the CSR-FP relationship, allowing the findings of this study to better inform the marketing strategies of luxury fashion brands.
3. Methodology
The purpose of this research was to examine how MI moderates the relationship between CSR and FP in the luxury fashion context. The methodological approach of this study involved gathering quantitative CSR and FP data from the Fashion Transparency Index and Capital IQ S&P 500 Database, respectively, constructing a cross-sectional panel dataset, and performing multivariable regression analysis to assess the statistical significance of the hypothesizes relationship between variables. Although incorporating quantitative measures of CSR into the methodology could provide the narrative depth that quantitative measures lack, the qualitative approaches employed in previous research, such as arranging for interviews with executives of luxury fashion brands (Bravo González, 2017), were beyond the scope of this study due to time constraints. Additionally, organizing data into a cross-sectional panel dataset, which captures multiple variables across multiple time intervals, enables the alignment of FP and control data from one year with CSR and MI data from the preceding year. This time-lag accounts for the divergence between the implementation of CSR initiatives and their impact on FP (Beqiraj).
Due to the longitudinal nature of the CSR and FP data in question, multiple regression analysis, which allows for time lag analysis, was the method used to evaluate the CSR-FP relationship. Moreover, unlike linear regression models, multiple regression models allow for the inclusion of multiple explanatory variables in analysis, such as control and moderating variables, aligning with the objectives of this study.
This study analyzed CSR and FP data of 12 luxury fashion brands from 2018-2019 and 2021-2022. The explanatory variables analyzed in this study included each of the five subsection scores (PC, G, T, KSF, and SI) and final score (FS) assigned to each brand of interest in the 2018 and 2021 Fashion Transparency Index reports, the control variables of leverage (LEV), inventory turnover (IT), and sales growth (SG) from 2019 and 2022, and the moderating variable of MI from 2018 and 2021. Financial data from the years 2019 and 2022, measured by two accounting-based metrics, return on assets (ROA), and return on equity (ROE), served as the response variables in this study.
3.1. Brand and Time Frame Selection
The sample and tie frame of this study were largely determined by the databases used to gather CSR and FP data. CSR data was extracted from the annual Fashion Transparency Index (FTI) reports, restricting the sample to brands consistently reviewed in the FTI reports. FP data was sourced from the Capital IQ S&P 500 database, which only provided fiscal reports spanning 2018-2022 for most luxury fashion companies at the time of data collection. This study was therefore limited to financial data from 2018-2022 and the corresponding FTI reports from 2017-2022 to enable time-lag analysis.
In this study, luxury fashion houses are defined as brands showing at each of the four major fashion weeks (London, Milan, New York, and Paris). Therefore, brands which simultaneously appear in the Vogue Runway Collection, which provides a comprehensive catalog of brand showing at the major fashion weeks, and in the FTI reports from 2017-2021 were initially included in this study.
14 brands fit this initial criterion: Bottega Veneta, Burberry, Chanel, Coach, Dior, Giorgio Armani, Gucci, Hermes, Louis Vuitton, Michael Kors, Miu Miu, Prada, Tommy Hilfiger, and Saint Laurent. However, changing the time frame from 2017-2021 to 2018-2022 for CSR data allowed for the inclusion of 4 additional brands—Dolce & Gabbana, Fendi, Valentino, and Versace—increasing the sample size by more than 25%. Therefore, the time frame of this study was altered to exclude CSR data from 2017. Later CSR data from 2019 and 2020 and the corresponding FP data from 2020 and 2021 were omitted to control for the effects of the COVID-19 pandemic on brand performance.
Brands with missing financial data for both 2019 and 2022—Coach, Chanel, Michael Kors, Miu Miu, Prada, and Tommy Hilfiger—were excluded from this study, although in consistency with (Beqiraj, 2021), brands missing one year of data were not omitted from the sample in order to avoid bias. Therefore, the final sample consisted of 12 brands (Bottega Veneta, Burberry, Dior, Dolce & Gabbana, Fendi, Giorgio Armani, Gucci, Hermes, Louis Vuitton, Saint Laurent, Valentino, and Versace), CSR data from 2018 and 2021, and corresponding financial data from 2019 and 2022.
3.2. Measuring CSR
CSR measurement was modeled after that of (Beqiraj, 2021), as the study’s examination of the CSR-FP relationship in the fashion industry closely aligned with the focus of this research. Given that CSR implementation varies amongst industries, it is important to measure CSR using an industry-specific database (Beqiraj, 2021). Therefore, CSR data was obtained from the FTI, which annually analyzes and ranks the 250 largest brands in the fashion industry based on public disclosure of CSR policies and practices in five areas: Policy & Commitments (PC), Governance (G), Traceability (T), Know-Show-Fix (KSF), and Spotlight Issues (SI). Section scores are calculated as a percentage of points received out of the total points possible. An overview of the criteria for each section is detailed in Table 1 below.
Table 1. Criteria for fashion transparency index dimensions.
| Policy & Commitments (PC) |
-social & environmental policies for employees and supply chain workers-implementation of policies-disclosure of annual targets and future relevant goals |
| Governance (G) |
-accountability of executive board for social and
environmental performance-implementation of oversight |
| Traceability (T) |
-disclosure of manufacturing, processing facilities & mills, and raw materials-availability of supplier information (ex. Address, gender breakdown, union representation, etc.) |
| Know-Show-Fix (KSF) |
-disclosure of due diligence processes-response to supplier ethical violations-reaction to worker complaints & ability of workers to file complaints |
| Spotlight Issues (SI) |
-disclosure of policies targeting the most urgent and
difficult problems facing the fashion industry |
Source: Fashion Transparency Index 2018.
Given that CSR is a multidimensional concept, and that the impact of one dimension may offset that of another (Nollet, Filis, & Mitrokostas, 2016), it is important to disaggregate the CSR variable into its section-specific components. Consequently, each section’s relationship with FP was tested for statistical significance independently. Once obtained, each brand’s section-specific scores from 2018 and 2021 were added to the study’s cross-sectional panel dataset in Excel.
3.3. Measuring Financial Performance
The majority of luxury fashion houses are either privately owned or subsidiaries of larger holding companies, neither of which are required to publicly publish unconsolidated financial reports. Therefore, it was necessary to obtain second-party FP data from the Capital IQ S&P 500 database, which contains financial records of public and private companies spanning five years. Financial records for most luxury brands were only available from 2019 to 2022 at the time of data collection, limiting the time frame of the study.
In existing CSR literature, FP is predominantly measured using either accounting-based metrics or market-based indicators. Although accounting-based measures are standardized, allowing for greater comparability across companies, market-based metrics reflect shareholder perceptions of a firm’s financial prospects, potentially indicating external validation of CSR initiatives. On the other hand, market valuations are not exact measures of a company’s FP, but rather a reflection of shareholders’ confidence in a firm.
Due to the lack of available market valuation data on private companies and subsidiaries, which make up the majority of luxury fashion brands, this study was unable to account for market-based indicators in analysis. Instead, two accounting-based metrics, return on assets (ROA) and return on equity (ROE), were used as proxies for FP.
ROA, calculated by dividing annual net income by average total assets, gauges the ability of a company to generate profit relative to its assets. ROE, which evaluates the ability of a firm to generate profit relative to its shareholder’s equity, is calculated by dividing annual net income by shareholder’s equity. Therefore, both metrics offer unique insights into the relationship between CSR and FP. Precalculated percentages for both ROA and ROE from 2019 and 2022 were extracted from the Capital IQ S&P 500 database for each brand and added to the study’s cross-sectional panel dataset in Excel.
3.4. Measuring Marketing Intensity & Other Control Variables
Data on MI, leverage (LEV), inventory turnover (IT), and sales growth (SG) was obtained from the Capital IQ S&P 500 database alongside ROA and ROE figures to minimize variability. In this study, MI data for 2018 and 2021 is calculated by dividing SG&A expenses by sales, representing the percentage of revenue allocated to SG&A activities.
Initially, MI was intended to be measured as the ratio of advertisement expenses to sales, as advertisement expenses directly reflect the amount of revenue allocated to marketing activities. In contrast, the SG&A figure includes general and administrative expenses in addition to marketing expenditure. However, given that most companies do not disclose advertisement expenses, prior research supports the use of SG&A expenses, which are more frequently reported, as a proxy for promotional expenses (Oh et al., 2014).
In addition to collecting MI data from 2018 and 2021, the MI values were multiplied by each CSR dimension (PC, G, T, KSF, and SI) to create the interaction terms MI*PC, MI*G, MI*T, MI*KSF, and MI*SI. Unlike the statistical significance of MI in the regression analysis, which determines whether MI independently correlates with FP, the significance of the aforementioned interaction terms—referred to collectively as MI*CSR in this paper—examines whether MI moderates the relationship between each CSR dimension and FP.
This study included LEV, IT, and SG as control variables to account for variations in firm-specific characteristics and economic conditions which may affect FP (Beqiraj, 2021). LEV, which represents the proportion of assets funded through debt, was calculated by dividing total liabilities by total assets. IT, which reflects the rate of inventory replenishment relative to cost of sales, was precalculated and thus obtained directly from the database. SG was calculated by subtracting the previous year’s sales from the current year’s sales, then dividing by the previous year’s sales. Once collected, all data was integrated into the study’s cross-sectional panel dataset in Excel.
3.5. Performing Data Analysis
All data analysis was performed through Stapplets, an online statistical analysis platform. IN order to achieve the study’s goal of evaluating the statistical significance of MI in moderating the CSR-FP relationship and the CSR-FP relationship itself in the luxury context, 20 regression models were developed using the platform’s multiple regression tool. Of the 20 models, 5 measured the impact of CSR on ROA without MI, 5 measured the impact of CSR on ROA with MI, 5 measured the impact of CSR on ROE without MI, and 5 measured the impact of CSR on ROE with MI. This breakdown allowed for the disaggregation of the CSR variable into its 5 dimensions—PC, G, T, KSF, and SI—for each category of regression models. A generic regression model from each of the 4 categories is shown below Table 2, and in actual analysis, the CSR variable is substituted for each of the 5 aforementioned dimensions.
Table 2. Generic regression models.
| ROA WITHOUT MI |
ROE WITHOUT MI |
| ROA = Constant + βCSR + βLEV + βIT + βSG |
ROE = Constant + βCSR + βLEV + βIT + βSG |
| ROA WITH MI |
ROE WITH MI |
| ROA = Constant + βCSR + βLEV + βIT + βSG + βMI + βMI*CSR |
ROE = Constant + βCSR + βLEV + βIT + βSG + βMI + βMI*CSR |
Source: Author’s Own Models.
Once formulated, two statistical measures, R2 Value and p-value, were used to analyze the regression models. The R2 value of a regression model measures the proportion of variance in the response variable that can be explained by the explanatory variables. To investigate H3, the R2 values of the models which included MI as an explanatory variable were analyzed in comparison to those which did not in order to determine whether accounting for MI improves the accuracy of the regression models in predicting FP.
The p-value of each explanatory variable, which determines the probability that any observed differences between an explanatory and response variable are due to chance, was used for two purposes. First, the value was used to determine the statistical significance of each explanatory CSR variable’s correlation with the response variable in the presence of MI to assess H1. The Pearson’s Correlation Coefficient was then used to determine the strength and direction of each statistically significant explanatory variable’s correlation with FP.
Second, in order to investigate H2, the p-values of the interaction terms MI*PC, MI*G, MI*T, MI*KSF, and MI*SI, which represent the combined effect of CSR and MI on FP, were measured to determine whether MI plays a statistically significant role in moderating the relationship between CSR and FP. The Pearson Correlation Coefficients of the statistically significant interaction terms were then used to measure the strength and direction of the role MI plays in moderating the relationship between CSR and FP.
4. Results, Analysis & Conclusion
4.1. Descriptive Statistics & Overall Results
Table 3 presents the descriptive statistics of the main variables investigated in this study. As mentioned under the methodology section, in consistency with (Beqiraj, 2021), brands missing one year of data were not excluded from the data set in order to avoid bias. The observation count of 20 instead of 24 reflects the exclusion of Gucci, Saint Laurent, and Versace from the 2018-2019 dataset and Valentino from the 2021-2022 dataset due to missing data.
Table 3. Descriptive statistics of main variables.
|
Minimum |
Maximum |
Mean |
Standard Dev. |
No. of
Observations |
| PC |
0 |
0.98 |
0.62 |
0.31 |
20 |
| G |
0 |
0.92 |
0.48 |
0.34 |
20 |
| T |
0 |
0.66 |
0.09 |
0.17 |
20 |
| KSF |
0 |
0.45 |
0.18 |
0.16 |
20 |
| SI |
0 |
0.66 |
0.19 |
0.19 |
20 |
| MI |
0.001 |
0.64 |
0.27 |
0.21 |
20 |
| ROA |
−0.04 |
0.3 |
0.09 |
0.08 |
20 |
| ROE |
−0.162 |
0.69 |
0.2 |
0.2 |
20 |
Source: Author’s Own Calculations.
Table 4 displays the results of the 20 multiple regression models outlined in the methodology section. The table provides the Pearson’s Correlation Coefficient (Coef), P-Values, T-Values, and SE values of the CSR dimensions and MI*CSR interaction terms of each model (the MI*CSR term is only present in models which tested MI as a moderating variable). Additionally, the table provides the R2 value of each model. The remainder of the results section will be divided into a discussion of the p-values and coefficients of the CSR dimensions and MI*CSR terms of each regression model, as well as a discussion of the R2 values of each of the regression models. This information, which is provided in Table 4, is repeated in smaller tables for the reader to reference in each specific results/discussion section.
Table 4. Coefficients, P-Values, SE Values, T-Values, and R2 Values of multiple regression models.
| ROA WITHOUT MI |
ROE WITHOUT MI |
|
Coef |
P |
R2 |
SE |
T |
|
Coef |
P |
R2 |
SE |
T |
| PC |
0.038 |
0.676 |
0.634 |
0.089 |
0.427 |
PC |
0.06 |
0.796 |
0.561 |
0.229 |
0.263 |
| G |
0.039 |
0.613 |
0.636 |
0.076 |
0.516 |
G |
0.094 |
0.637 |
0.566 |
0.195 |
0.481 |
| T |
−0.067 |
0.609 |
0.636 |
0.129 |
−0.522 |
T |
−0.449 |
0.17 |
0.613 |
0.311 |
−1.443 |
| KSF |
−0.075 |
0.648 |
0.635 |
0.16 |
−0.466 |
KSF |
−0.254 |
0.543 |
0.57 |
0.408 |
−0.623 |
| SI |
−0.146 |
0.283 |
0.658 |
0.131 |
−1.114 |
SI |
−0.445 |
0.198 |
0.607 |
0.33 |
−1.348 |
| ROA WITH MI |
ROE WITH MI |
|
Coef |
P |
R2 |
SE |
T |
|
Coef |
P |
R2 |
SE |
T |
| PC |
0.189 |
0.374 |
0.652 |
0.205 |
0.92 |
PC |
0.602 |
0.263 |
0.604 |
0.514 |
1.171 |
| MI*PC |
−0.328 |
0.44 |
0.652 |
0.412 |
−0.796 |
MI*PC |
−1.209 |
0.262 |
0.604 |
1.032 |
−1.172 |
| G |
0.085 |
0.575 |
0.64 |
0.147 |
0.575 |
G |
0.348 |
0.364 |
0.587 |
0.37 |
0.941 |
| MI*G |
−0.128 |
0.732 |
0.64 |
0.365 |
−0.35 |
MI*G |
−0.733 |
0.439 |
0.587 |
0.918 |
−0.798 |
| T |
−1.971 |
0.013 |
0.773 |
0.69 |
−2.858 |
T |
−4.602 |
0.021 |
0.731 |
1.761 |
−2.613 |
| MI*T |
7.599 |
0.015 |
0.773 |
2.721 |
2.793 |
MI*T |
16.549 |
0.033 |
0.731 |
6.95 |
2.381 |
| KSF |
−0.19 |
0.533 |
0.641 |
0.297 |
−0.641 |
KSF |
−0.417 |
0.593 |
0.573 |
0.761 |
−0.548 |
| MI*KSF |
0.34 |
0.696 |
0.641 |
0.851 |
0.4 |
MI*KSF |
0.373 |
0.867 |
0.573 |
2.178 |
0.171 |
| SI |
−0.511 |
0.061 |
0.721 |
0.249 |
−2.052 |
SI |
−1.321 |
0.057 |
0.673 |
0.632 |
−2.089 |
| MI*SI |
0.964 |
0.153 |
0.721 |
0.635 |
1.519 |
MI*SI |
2.257 |
0.185 |
0.673 |
1.612 |
1.4 |
Source: Author’s Own Calculations.
4.2. Results: P-Values and Coefficients of CSR Dimensions
Table 5 displays the p-values and coefficients of each CSR dimension modeled in the regression analysis. For example, referring to Table 5, PC had a p-value of 0.676 and a coefficient of 0.038023 in the regression model which examined the impact of the PC on ROA without the moderating role of MI. Similarly, in the regression model which examined the impact of PC on ROE with the moderating role MI, PC had a p-value of 0.263 and a coefficient of 0.60187. In this study, p-values less than 0.05 were considered statistically significant; all statistically significant values are highlighted in Table 5.
Table 5. P-values and coefficients of CSR dimensions.
| ROA WITHOUT MI |
ROE WITHOUT MI |
| PC p-value: 0.676, coefficient: 0.038032 |
PC p-value: 0.796, coefficient: 0.060315 |
| G p-value: 0.613, coefficient: 0.039231 |
G p-value: 0.637, coefficient: 0.093843 |
| T p-value: 0.609, coefficient: −0.067101 |
T p-value: 0.17, coefficient: −0.4491 |
| KSF p-value: 0.648, coefficient: −0.074581 |
KSF p-value: 0.543, coefficient: −0.253837 |
| SI p-value: 0.061, coefficient: −0.146122 |
SI p-value: 0.198, coefficient: −0.445226 |
| ROA WITH MI |
ROE WITH MI |
| PC p-value: 0.374, coefficient: 0.188893 |
PC p-value: 0.263, coefficient: 0.60187 |
| G p-value: 0.575, coefficient: 0.084532 |
G p-value: 0.364, coefficient: 0.347821 |
| T p-value: 0.013, coefficient: −1.97092 |
T p-value: 0.021, coefficient: −4.602144 |
| KSF p-value: 0.533, coefficient: 0.339829 |
KSF p-value: 0.593, coefficient: −0.416889 |
| SI p-value: 0.061, coefficient: −0.510745 |
SI p-value: 0.057, coefficient: −1.320776 |
Source: Author’s Own Calculations.
The p-values of all 5 CSR dimensions in the regression models exploring the CSR-FP correlation without MI were greater than 0.05, and thus considered statistically insignificant. Additionally, the p-values of PC, G, KSF, and SI were greater than 0.05 in the regression models which included MI as a moderating variable, and therefore also considered insignificant. However, the p-values of T were statistically significant in both regression models which included MI, with p-values of 0.013 for the ROA model and 0.021 for the ROE model. Therefore, the T coefficients of −1.97092 and −4.602144 for the MI inclusive ROA and ROE models suggest a negative correlation between T and FP in the moderating presence of MI.
Table 6. P-values and coefficients of CSR interaction terms.
| ROA WITH MI |
ROE WITH MI |
| MI*PC p-value: 0.44, coefficient: −0.328012 |
MI*PC p-value: 0.262, coefficient: −1.209394 |
| MI*G p-value: 0.732, coefficient: −0.127912 |
MI*G p-value: 0.439, coefficient: −0.732955 |
| MI*T p-value: 0.015, coefficient: 7.598997 |
MI*T p-value: 0.033, coefficient: 16.549492 |
| MI*KSF p-value: 0.696, coefficient: 0.339829 |
MI*KSF p-value: 0.867, coefficient: 0.372555 |
| MI*SI p-value: 0.153, coefficient: 0.963878 |
MI*SI p-value: 0.185, coefficient: 2.257019 |
Source: Author’s Own Calculations.
Table 6 presents the p-values and coefficients of the interaction terms between MI and CSR dimensions in the regression models assessing the influence of CSR on ROA and ROE in the moderating presence of MI. The p-values of the interaction terms MI*PC, MI*G, MI*KSF, and MI*SI were all less than 0.05, and thus statistically insignificant. However, the p-values of MI*T were 0.015 and 0.033 for ROA and ORE, indicating statistical significance. Therefore, the MI*T coefficients of 7.598997 and 16.549492 for ROA and ROE suggest that MI positively moderates the relationship between T and FP.
4.3. Discussion: P-values and Coefficients of CSR Dimensions
The insignificance of the p-values of PC, G, KSF, and SI in the moderating presence of MI indicate that there is no correlation between these areas of CSR and FP. Thus, H1, which posited that all CSR dimensions would significantly correlate with FP is rejected. Additionally, the insignificance of the MI*PC, MI*G, MI*KSF, and MI*SI interaction terms indicate that MI does not influence the correlation between PC, G, KSF, and SI, and therefore H2, which posited that MI would significantly moderate the relationship between all CSR dimensions and FP is rejected. The lack of correlation between the PC, G, KSF, and SI and their respective interaction terms and FP suggests that luxury fashion companies do not financially benefit from allocating resources towards implementing and promoting CSR policies which fall into the aforementioned categories.
Therefore, designer brands should refrain from promoting PC, G, KSF, and SI initiatives to optimize FP. However, the statistical significance of both T and the interaction term MI*T in the moderating presence of MI indicates that traceability-oriented CSR initiatives impact the FP of luxury fashion brands. Although T has a standalone negative correlation with FP, which would initially suggest that traceability-oriented policies negatively impact FP, the interaction term MI*T is significantly positive. This indicates that higher levels of MI weaken the negative correlation between T and FP, and therefore luxury fashion companies may financially benefit from allocating a greater proportion of marketing expenditure towards traceability-oriented initiatives to enhance FP.
4.4. Results: R2 Values of Regression Models
Table 7 showcases the R2 values for the 20 multiple regression models developed within the study. For instance, the R2 value of the multiple regression model analyzing the impact of PC on ROA without the moderating influence of MI is 0.634. Similarly, the R2 value of the regression model assessing the impact of PC on ROA with the moderating role of MI is 0.652. Although there is no standardized threshold of statistical significance for R2 values, the higher the R2 value of a model, the more accurate the model is in predicting the influence of the explanatory variables on the response variable.
Therefore, this study examined whether the R2 value of each regression model without MI increased when MI was included as a moderating variable. For example, the R2 value for the model examining the impact of PC on ROA without MI was compared to the model examining the influence of PC on ROA with MI. The models which increased in R2 value are highlighted in Table 7. Referring to Table 7, it is evident that the R2 value of every model increased when MI was included as a moderating variable.
Table 7. R2 values of regression models.
| ROA WITHOUT MI |
ROE WITHOUT MI |
| PC R2: 0.634 |
PC R2: 0.561 |
| G R2: 0.636 |
G R2: 0.566 |
| T R2: 0.636 |
T R2: 0.613 |
| KSF R2: 0.635 |
KSF R2: 0.57 |
| SI R2: 0.658 |
SI R2: 0.607 |
| ROA WITH MI |
ROE WITH MI |
| PC R2: 0.652 |
PC R2: 0.604 |
| G R2: 0.64 |
G R2: 0.587 |
| T R2: 0.773 |
T R2: 0.731 |
| KSF R2: 0.641 |
KSF R2: 0.573 |
| SI R2: 0.721 |
SI R2: 0.634 |
Source: Author’s Own Calculations.
4.5. Discussion: R2 Values of Regression Models
The increased R2 values of the regression models which included MI as a moderating variable indicates that including MI as a moderating variable in modeling the CSR-FP relationship improves the accuracy of the regression models in predicting FP. Therefore, H3, which suggested that including MI as a moderating variable will improve the accuracy of the regression analysis in predicting the CSR-FP relationship, is accepted. Based on these findings, future research should include MI in regression analysis when modeling the CSR-FP relationship in the luxury fashion context.
5. Limitations & Directions for Future Research
The findings of this research are limited due to several methodological restrictions. For one, this study only examined data from 12 luxury fashion brands from 2018-2019 and 2021-2022. Therefore, given the relatively small dataset of this study, the regression analysis is limited in its ability to accurately predict the CSR-FP relationship. Additionally, the timeline of this project prevented the utilization of advanced statistical analysis techniques employed in other studies, such as robustness testing (Beqiraj, 2021), which may have enhanced the validity of the findings given the sample size and time frame restrictions. Future researchers should therefore replicate this study using a larger sample of luxury fashion brands and examine data over a longer period.
Secondly, all data in this study was collected from the FTI and Capital IQ S&P 500 databases. Although the use of databases minimizes variability in CSR and FP measurement, allowing for more accurate comparison across brands, the scope of this study was limited to the information available through the FTI and Capital IQ S&P 500 databases as a result. For instance, although previous research indicates that further disaggregating CSR into its internal and external dimensions may increase the validity of the CSR-FP correlation (Gordano & Chiaudano, 2021), the FTI does not allow for the further disaggregation of its 5 sections, which group internal and external CSR initiatives together. As a result, this study was unable to account for the internal-external CSR variance in analysis.
Similarly, while advertising expenditure is a more precise measure of MI than the aggregate SG&A expenditure figure, the Capital IQ S&P 500 database did not report advertising expenditure for most private companies and subsidiaries. Therefore, SG&A expenditure, which includes administrative expenses unrelated to promotional activities, was used to measure MI in this study. Researchers exploring the role of MI in moderating the CSR-FP relationship should thus consider obtaining data beyond databases, which would allow for the further disaggregation of the CSR and MI variables in analysis.
Additionally, previous research suggests accounting-based FP measures are subject to manipulation through managerial accounting tactics (Ahamed et al., 2014), and therefore suggest incorporating both accounting-based and market-based FP metrics in FP analysis. However, because most luxury fashion brands are privately owned and require access to proprietary financial information to calculate market value, this study used accounting-based FP metrics in analysis.
Therefore, future studies should consider obtaining the FP data necessary to incorporate both market-based and accounting-based measures in analysis. In addition, thus study was unable to include certain control variables recommended by prior studies, such as R&D expenses (Beqiraj, 2021), as the data was not consistently available for every luxury fashion brand sampled in the Capital IQ S&P 500 database. Incorporating such controls in the future may enhance the accuracy of regression analysis in modeling the CSR-FP relationship in the moderating presence of MI.
Furthermore, the researcher’s limited statistical background prevented the utilization of advanced statistical analysis tests, such as robustness checks, the Hausman test, and fixed & random effects, which may have enhanced the validity of the data. It is therefore recommended that future studies which replicate the methodology of this study incorporate the aforementioned statistical analysis techniques to enhance the validity of future findings.
6. Conclusion
Based on the multivariable regression analysis, it can be concluded that CSR initiatives within the PC, G, KSF, and SI dimensions have no correlation with the FP of luxury fashion brands. Despite this finding, CSR initiatives within the aforementioned categories may still influence FP. As professor of marketing at Kyungpook National University Hannah Oh explains, “highly visible firms are expected to engage rigorously in various CSR activities, so their actual engagement in CSR is not rewarded” (Oh et al., 2014). Therefore, given the high visibility of the luxury fashion industry, the lack of correlation between the majority of CSR dimensions and FP can be attributed to baseline stakeholder expectations of designer brands.
However, based on the findings of this study, luxury fashion brands should refrain from pursuing CSR goals in the aforementioned areas until a definitive relationship is proven using a larger dataset. The statistical significance of both the independent traceability dimension and the MI*T interaction term in regression analysis suggests that traceability-oriented CSR initiatives impact the FP of luxury fashion brands. Therefore, luxury fashion companies may benefit from allocating a greater proportion of marketing funds towards implementing traceability-oriented CSR initiatives, such as supply chain transparency policies. By investigating the role of MI in moderating the CSR-FP relationship using industry-specific, disaggregated measures of CSR in the luxury fashion context, this study contributes valuable insights to existing CSR literature.
The findings of this study add to the ongoing discussion of the role of CSR in the luxury fashion industry by providing evidence to support the inclusion of MI in future analysis. Additionally, this research informs luxury fashion brands of the optimal CSR strategies to implement given the unique characteristics of the luxury fashion industry, allowing designer brands to optimize FP while simultaneously benefiting stakeholders.