Buffering Effect of CSR Reputation During Product Recalls: Evidence from Global Automakers Across Institutional Contexts

Abstract

Multinational corporations (MNCs) face significant reputational and performance risks from product recalls, yet the severity of these consequences varies across national markets. While prior research suggests that corporate social responsibility (CSR) can buffer against such crises, limited attention has been paid to how country-level institutions shape this effect. This study examines whether—and under what institutional conditions—CSR reputation mitigates the negative market consequences of product recalls. We focus on how the insurance-like effect of CSR varies with the level of corruption in a country’s institutional environment. Using panel regression analysis and hand-collected data from 14 global automotive manufacturers across eight countries (2007–2015), we find that firms with stronger CSR reputations experience significantly smaller declines in market share after recall announcements. Furthermore, this buffering effect is amplified in countries with higher corruption levels, suggesting that when formal institutional trust is weak, CSR signals play a greater role in stakeholder perceptions. These findings advance CSR literature by showing that its reputational benefits are contingent on institutional context and contribute to international business scholarship by revealing how national-level corruption interacts with firm-level reputational assets during crises.

1. Introduction

In September 2015, the global automotive industry was rocked by revelations that Volkswagen had deliberately manipulated emissions tests on 11 million diesel vehicles worldwide. The scandal instantly wiped nearly USD 30 billion from the company’s market value [1] and triggered one of the largest product recalls in automotive history. Despite the global scope of the incident, market responses varied widely across countries—ranging from devastating sales collapses in some markets to relatively muted reactions in others [2]. This striking divergence is not unique to Volkswagen. Similar discrepancies emerged during Toyota’s unintended acceleration recalls [3,4], GM’s ignition switch failures [4], and Honda’s exposure to the Takata airbag crisis. In each of these cases, consumer reactions and reputational consequences differed markedly across national markets.

These real-world patterns raise the following questions: Why are some firms able to retain consumer loyalty and stabilize market share in the midst of severe product crises, while others experience immediate abandonment? And what explains the stark variation in these effects across different institutional environments? While the existing research has extensively documented the negative consequences of product recalls on firm performance [5], the institutional conditions under which these effects unfold—and the reasons why they differ—remain insufficiently theorized and empirically examined.

Corporate social responsibility (CSR) reputation—defined as stakeholders’ collective perceptions of a firm’s commitment to environmental, social, and governance principles beyond legal requirements—has emerged as one potential buffer against crisis-related damage, with scholars theorizing that positive CSR reputations may provide “insurance-like” protection during controversies [6,7]. This insurance-like effect suggests that firms with strong CSR credentials have accumulated moral capital and stakeholder goodwill, which can be drawn upon during crises to mitigate negative stakeholder assessments. Empirical evidence has largely supported this buffering effect [8,9].

However, a critical question that remains unanswered is the extent to which the protective value of CSR during crises varies across different institutional environments, specifically the formal and informal rules, norms, and enforcement mechanisms that govern economic activity within a society. This question is particularly important for multinational corporations (MNCs) operating in diverse institutional contexts. Notably, in environments characterized by high levels of corruption—where formal institutions lack credibility and stakeholders harbor deep skepticism toward both government and corporate actors—the dynamics of trust, attribution, and forgiveness during corporate crises may function very differently than in low-corruption contexts.

Indeed, Rodriguez et al. (2006) provided a valuable perspective by highlighting how corruption, CSR, and corporate engagement are interdependent elements of the non-market environment confronting multinational enterprises (MNEs) [8]. Their integrative framework emphasizes that institutional variation across countries fundamentally alters how CSR is interpreted and deployed. For instance, in highly corrupt environments—where formal institutions are weak and societal trust in governance is eroded, CSR may function not only as a moral stance but also as a strategic response to legitimacy deficits. Ucar and Staer (2020) offered empirical evidence from a large panel of U.S. firms, showing that companies headquartered in more corrupt localities tend to engage in lower levels of CSR [9]. They argued that this occurs for the following two primary reasons: first, because widespread local misconduct reduces the social stigma of unethical behavior; and second, because stakeholders in high-corruption regions are more skeptical of CSR, perceiving it as insincere or instrumentally motivated. Their work provides important baseline insights into how corruption influences CSR engagement, yet it leaves unexplored how these dynamics shift during critical moments such as product recalls.

Building on their foundation, we direct attention to how CSR reputation functions specifically during corporate crises. We theorize that in environments characterized by high institutional corruption, the informational value of CSR as a trust signal is amplified, enhancing its protective role in the face of product-harm events. Thus, while their study examines steady-state CSR levels, ours complements it by revealing how corruption moderates CSR’s crisis-mitigating effects. Despite extensive research on both CSR and product recalls, our understanding of how institutional corruption moderates the relationship between CSR reputation and crisis outcomes remains remarkably limited. This gap is especially relevant for firms operating in emerging and transitional economies, where formal institutions often fail to command stakeholder trust. We address this gap by examining how country-level corruption moderates the insurance-like effect of CSR during product recalls. This shift in emphasis brings to the forefront the following broader but underexplored question: To what extent does institutional corruption shape not just whether firms engage in CSR, but how effectively CSR reputation mitigates damage during crises such as product recalls?

Corruption—understood as the abuse of public office for private gain and the systemic undermining of institutional integrity—represents a fundamental breakdown in institutional integrity, diminishing trust in formal institutions and creating environments where stakeholders must rely more heavily on alternative signals of trustworthiness [8,10,11,12]. We theorize that in contexts characterized by high corruption, stakeholders place greater emphasis on firm-specific signals like CSR reputation when making attributions about firms’ underlying motives and values during crises. Consequently, the insurance-like protection afforded by CSR reputation may be particularly valuable in these environments.

We test our hypotheses in the context of the global automotive industry for several theoretically grounded reasons. First, the automotive sector represents one of the world’s most globalized industries, with manufacturers operating across diverse institutional contexts while maintaining standardized safety and quality protocols. Second, automotive products pose significant safety risks to consumers, making product recalls particularly salient events that generate intense media scrutiny and consumer attention. Third, the industry’s high capital intensity and long product lifecycles create substantial reputational stakes, as consumer trust directly affects multi-year purchase decisions. Finally, the sector exhibits considerable variation in CSR practices and recall frequencies across manufacturers and markets, providing rich empirical terrain for testing our institutional moderation hypothesis. These characteristics make the automotive industry an ideal setting for examining how institutional corruption shapes the relationship between CSR reputation and crisis outcomes.

In our analysis, we use a hand-collected longitudinal dataset of 14 global automakers and their 27 brands across eight countries from 2007 to 2015. This period offers unique analytical advantages, as follows: It encompasses multiple watershed moments in automotive crisis management, including Toyota’s unintended acceleration crisis (2009–2010), GM’s ignition switch failures (2014), and Volkswagen’s emissions scandal (2015). These years marked a transformative era in global recall governance, characterized by heightened regulatory scrutiny, increased media attention, and evolving consumer expectations regarding corporate accountability. The period also provides substantial institutional variation across countries, allowing us to test how corruption levels moderate CSR’s protective effects. Our comprehensive dataset, meticulously assembled from national regulatory agencies and global ESG databases, captures the full arc of these critical events and their market consequences.

Our empirical analysis reveals that CSR reputation offers a buffering, insurance-like effect against the adverse consequences of product recalls. Crucially, this effect is significantly more pronounced in high-corruption environments. This interaction challenges the prevailing assumption that CSR’s protective value is context-invariant and instead highlights the institutional contingency of stakeholder trust and attribution processes. For scholars, our findings suggest the need for more nuanced theories of CSR’s crisis-mitigating functions, specifically ones that incorporate institutional context as a key moderator. For managers, the results offer a counterintuitive insight where strategic investments in CSR may be especially valuable in markets marked by high institutional corruption, precisely those settings often perceived as risk-laden or governance-deficient.

This study makes three key contributions. First, we extend the insurance-like view of CSR by identifying institutional corruption as a boundary condition that moderates CSR’s protective effect during crisis events. Second, we contribute to institutional theory and non-market strategy by showing how CSR substitutes for weak formal institutions in high-corruption settings, where stakeholder trust in public governance is limited. Third, we advance crisis management and attribution theory by demonstrating how stakeholder responses are shaped not only by firm-level behavior but also by the broader institutional environment in which a crisis occurs. In doing so, we offer a context-sensitive understanding of when and how CSR reputation serves as a strategic asset in global markets.

The remainder of the paper is organized as follows. Section 2 develops the theoretical framework and formulates the study’s hypotheses. Section 3 outlines the data sources, variable construction, and empirical strategy. Section 4 presents the main findings. Section 5 discusses the theoretical and managerial implications of our findings, outlines key limitations and directions for future research, and concludes the paper.

2. Theoretical Background and Hypotheses Development

2.1. The Insurance-like Effect of CSR During Product Recalls

Product recalls are among the most reputationally and financially damaging crises that firms can face, often triggering consumer backlash, erosion of brand equity, and significant financial losses [13]. These risks are especially salient in the automotive industry, where products are safety-critical, media exposure is high, and regulatory scrutiny is intense [14,15]. For multinational manufacturers, recall decisions are further complicated by cross-national variations in legal requirements, media norms, and stakeholder expectations.

To mitigate the negative consequences of such crises, firms have increasingly turned to CSR as a reputational shield. Shiu and Yang (2017) showed that CSR functions as a buffer, offering “insurance-like” protection during adverse events by generating moral capital [6,7]. Empirical studies offered broad support for this buffering effect. Flammer (2013) showed that firms with strong environmental performance suffer less market penalty following environmental controversies [16], while Klein and Dawar (2004) found that positive CSR associations reduce consumer blame during product-harm events [17]. Shiu and Yang (2017) demonstrated that CSR engagement attenuates negative stakeholder reactions following product recalls [7].

Research suggests that the protective value of CSR during crises operates through three interrelated mechanisms—moral capital accumulation, the halo effect on stakeholder attributions, and credible signaling [6,17,18,19]. These mechanisms work synergistically to buffer firms from the full impact of product recalls.

First, CSR facilitates the accumulation of moral capital by establishing a long-term track record of ethical behavior, stakeholder engagement, and public responsibility [6,7]. Moral capital represents the accumulated reservoir of stakeholder goodwill and trust that firms build through ethical conduct over time, which can be drawn upon during periods of reputational threat. When firms with robust CSR reputations disclose product defects or issue voluntary recalls, stakeholders are more likely to interpret these actions as responsible and principled rather than negligent or evasive, creating an intangible asset that serves as a protective buffer during adversity. When crises strike, this reservoir of trust enables firms to preserve their reputation and maintain stakeholder support. Stakeholders, in turn, may attribute more benign motives to socially responsible firms, interpreting crises as unfortunate events rather than signs of systemic failure or ethical lapse [6,7].

Second, CSR creates a ‘halo effect’ that influences stakeholder attributions during crises. According to attribution theory, individuals evaluate the causes of negative events by integrating organizational behavior with contextual cues [18]. When firms have established strong CSR records, stakeholders demonstrate what Klein and Dawar (2004) describe as ‘attributional bias in favor of the firm’ [17]. Specifically, stakeholders are more inclined to attribute crises at CSR-active firms to uncontrollable external factors or genuine errors, rather than to managerial irresponsibility or moral failure. This favorable attribution process reduces the reputational damage typically associated with crisis events.

Third, CSR serves as a credible signal of long-term orientation and stakeholder commitment, especially in environments of uncertainty and information asymmetry [19,20]. Signaling theory suggests that firms can convey unobservable qualities—such as integrity or future behavior—through observable actions. Sustained CSR investment signals to stakeholders that the firm values trust, accountability, and reputation more than short-term profit gains. Thus, even during adverse events like product recalls, firms with high CSR reputations are perceived as unlikely to compromise stakeholder interests, which helps preserve their legitimacy and consumer confidence.

Recent empirical evidence further supports this buffering effect in crisis contexts. Kang et al. [15] found that firms with stronger CSR records experience less negative market reaction to product recalls, particularly in high-severity cases. Similarly, Gao et al. [21] demonstrated that reputational capital accumulated through CSR helps firms withstand both institutional and reputational shocks.

Together, these mechanisms explain why CSR can serve as a reputational buffer during recall crises. They reduce perceived blame, affirm stakeholder trust, and frame corporate responses in a more favorable light. Therefore, we hypothesize the following:

Hypothesis 1 (H1): 

Firms with higher pre-recall CSR reputations will experience smaller declines in market share following automotive recalls compared to firms with lower CSR reputations.

2.2. Institutional Corruption and CSR’s Insurance Value

We now turn to how institutional context shapes CSR’s protective effect during product crises. Specifically, we examine the moderating role of country-level corruption.

Institutional corruption—the systemic abuse of public authority for private gain—constitutes a profound breakdown in institutional integrity [22]. In highly corrupt environments, formal institutions lose their credibility, creating a pervasive trust deficit that compels stakeholders to seek alternative indicators of corporate trustworthiness [8,9,10,23]. As a result, we posit that when firms face crises such as product recalls, stakeholders in these settings confront heightened information asymmetry and ambiguity, making them more sensitive to firm-specific trust signals that can reduce uncertainty and guide attribution [8].

Empirical research offers preliminary support for this reasoning. For instance, Ioannou and Serafeim (2012) found that the market consequences of CSR vary systematically across institutional environments [24], and El Ghoul et al. (2017) demonstrated that the financial value of CSR is amplified in countries with weaker market institutions [25].

Building on this foundation, we theorize that CSR activities function as particularly salient trust signals in corrupt environments. This effectiveness stems from several interrelated factors. First, CSR often constitutes voluntary, discretionary behavior that exceeds legal compliance [26]. In corruption-heavy institutional environments—where laws are weak or selectively enforced—such discretionary behavior is more visible and carries greater informational value. Matten and Moon (2008) argued that in countries where public governance is limited, CSR becomes a more prominent mechanism for firms to demonstrate moral legitimacy [27].

Second, CSR frequently addresses governance gaps in areas where public institutions underperform, such as environmental protection, infrastructure development, and community welfare [28,29]. In corrupt settings where governments fail to meet basic societal obligations, stakeholders increasingly look to firms to fill these institutional voids. Hence, CSR becomes strategically valuable in such contexts because it performs a substitutive role, enhancing firm legitimacy by compensating for institutional deficiencies [30].

Third, CSR often involves independent verification and public disclosure, providing relatively reliable information about firm behavior in environments where official data and state-regulated disclosures may be suspect [31]. Oikonomou et al. (2014) find that CSR performance is especially effective in mitigating risk in low-transparency markets, where reliable information is otherwise scarce [32].

Fourth, sustained CSR investment signals a firm’s long-term orientation and integrity, even in settings where opportunistic behavior is prevalent and rarely sanctioned [11]. In such environments, CSR serves not only as a moral gesture but as a strategic differentiator, helping firms build a reputation as “honest actors” amid widespread distrust. Keig et al. (2015) showed that firms operating in more corrupt countries are subject to higher scrutiny, thereby increasing the signaling value of positive distinguishing characteristics such as CSR engagement [33].

Research on attribution theory further suggests that stakeholders in corruption-prone environments are likely to rely more heavily on firm-specific cues like CSR when interpreting the causes and meaning of crisis events [34]. A voluntary recall issued by a firm with a strong CSR reputation may thus be interpreted not as a sign of failure, but as a rare instance of ethical commitment. Coombs (2007) emphasizes that prior reputation significantly shapes stakeholder responses to crisis communication, with stronger reputations eliciting greater benefit of the doubt [35].

These arguments collectively suggest that institutional corruption amplifies CSR’s insurance-like effect. In environments where actors are commonly perceived as untrustworthy, a history of socially responsible behavior provides a scarce but credible signal of firm integrity. Facing unreliable institutional cues, stakeholders rely more heavily on CSR as an indicator of which firms deserve continued trust during product failures. Based on this reasoning, we propose the following hypothesis:

Hypothesis 2 (H2): 

The buffering effect of CSR reputation on the recall–market share relationship is stronger in countries with higher levels of corruption.

3. Methodology

3.1. Research Context

Our research is situated in the global automotive industry—a context that offers a rich setting for examining how product recalls interact with firms’ corporate social responsibility (CSR) reputations across institutional environments. Automotive recalls are high-stakes events due to their direct implications for consumer safety, their regulatory visibility, and their reputational sensitivity [36]. Unlike low-salience product defects in other industries, automotive recalls often receive substantial media attention, invoke regulatory intervention, and provoke public scrutiny, making them an ideal setting through which to study stakeholder attribution and reputational risk. Moreover, the global automotive industry is characterized by large multinational corporations (MNCs) that operate across highly diverse institutional contexts. This variation in national governance, regulatory enforcement, and corruption levels presents a valuable opportunity to investigate how institutional conditions shape the reputational buffering effects of CSR.

3.2. Data Sources and Sample Construction

To empirically test our hypotheses, we assembled a unique panel dataset comprising 14 global automotive manufacturers and their 27 associated brands (or “makes”) across eight national markets over a nine-year period (2007–2015).

The selected markets—the United States, South Korea, Japan, China, the United Kingdom, Germany, Canada, and Australia—were chosen based on their institutional diversity, data availability, and market relevance. These countries represent a wide spectrum of institutional environments, particularly in terms of corruption levels.

Our sample period spans 2007–2015, a timeframe selected for both theoretical and empirical considerations. Methodologically, this period provides several distinct advantages for testing our hypotheses. First, it includes sufficient pre- and post-crisis observations for multiple major recall events, enabling robust difference-in-differences analyses. Second, the period captures significant variation in both recall severity and CSR investments across firms, essential for identifying our moderating effects. Third, these years witnessed substantial cross-country variation in corruption indices, allowing us to test our institutional moderation hypothesis with meaningful variance in our key contextual variable.

From a data availability perspective, this timeframe represents the most comprehensive period for which standardized recall data could be collected across all eight countries in our sample. National regulatory databases maintained consistent reporting standards during these years, while post-2015 data collection faced challenges due to evolving disclosure requirements and database migrations in several jurisdictions. Additionally, our ESG data sources provide complete and comparable CSR metrics for all sample firms throughout this period, whereas earlier years lack consistent sustainability reporting and later years introduced new measurement frameworks that would compromise comparability.

Data were collected from multiple authoritative sources and integrated at the make–country–year level. Ward’s AutoWorld was used to identify each firm’s makes and to collect sales data; product recall data were compiled directly from national regulatory agencies; CSR reputation scores were obtained from the Refinitiv ESG database; financial data were sourced from Compustat Global; and country-level corruption indicators were drawn from the World Bank’s Worldwide Governance Indicators (WGI) and the Bayesian Corruption Indicators (BCI), both housed within the Quality of Government Institute’s datasets. After removing observations with missing or unmatched variables, the final dataset used for hypothesis testing included 1192 make–country–year observations.

3.3. Variable Measurement

3.3.1. Dependent Variable

Our dependent variable is the annual market share of each automobile make in each country–year. Data on vehicle sales by make and country were obtained from Ward’s AutoWorld, a widely used and authoritative source for global automotive market data. For each observation, the market share was calculated as the number of vehicles sold by a given make in a specific country and year divided by the total number of vehicles sold in that country and year. To facilitate interpretation and align the scale with standard reporting formats, we rescaled the resulting market share values by multiplying them by 100, such that all values would be expressed in percentage points rather than proportions.

3.3.2. Independent Variable: Recall Intensity

Our focal independent variable is recall intensity, measured as the annual number of product recalls issued for each make. To construct this variable, we manually collected recall data from the websites of official government agencies responsible for automobile safety and recall disclosure. Specifically, recall data were obtained from official national regulatory agencies including the National Highway Traffic Safety Administration (NHTSA) for the United States, the Korea Transportation Safety Authority (KTSA) for South Korea, the Ministry of Land, Infrastructure, Transport, and Tourism (MLIT) for Japan, the State Administration for Market Regulation (SAMR) for China, the Driver and Vehicle Standards Agency (DVSA) for the United Kingdom, Kraftfahrt-Bundesamt (KBA) for Germany, Transport Canada for Canada, and the Australian Competition and Consumer Commission (ACCC) for Australia.

Each recall was matched to the relevant make and year and subsequently aggregated to construct a count of recall frequency at the make–year level. After this procedure, recall intensity was calculated as the natural logarithm of one plus the total number of recalls issued for a given make in a given year, to address the skewed distribution of recall frequency.

3.3.3. Moderating Variable: CSR Reputation

We measured each firm’s corporate social responsibility (CSR) reputation using the Environmental, Social, and Governance (ESG) Combined Score provided by the Refinitiv ESG database (formerly known as Thomson Reuters Asset4). This metric has become one of the most widely used standardized ESG indicators in recent CSR and sustainability research [37,38,39].

The ESG Combined Score is constructed by aggregating firm performance across the three ESG pillars based on publicly available disclosures (e.g., environmental policies, employee diversity, and governance structures), while also incorporating negative adjustments for ESG-related controversies—such as involvement in corruption, environmental damage, or labor rights violations. Firms with a clean controversy record retain an ESG Combined Score equivalent to their underlying ESG performance; firms associated with severe or frequent controversies receive downward adjustments [40,41].

The ESG scores were linked to the parent companies of each make in our sample (e.g., linking Toyota and Lexus to Toyota Motor Corporation). The use of firm-level CSR scores rather than make-level scores is appropriate because most CSR communications, reporting, and stakeholder attributions are made largely at the global corporate level.

This score is particularly suitable for capturing CSR reputation for two reasons. First, it reflects not only a firm’s proactive social and environmental investments but also its exposure to reputational risks, making it a more comprehensive indicator of stakeholder-facing CSR performance than purely policy-based measures. Second, because ESG controversies are typically more visible to external stakeholders than granular CSR inputs, the combined score better approximates how a firm’s CSR behavior is interpreted and evaluated by the public [42,43].

To align with the theoretical notion of reputational stock and to mitigate simultaneity concerns, we used the lagged (t − 1) ESG Combined Score, following standard practice in the literature [38,39]. The score was assigned at the firm level and matched to each make within the manufacturer’s brand portfolio (e.g., Toyota and Lexus receive Toyota Motor Corporation’s ESG score), reflecting the reality that most CSR communication and stakeholder attributions occur at the global corporate level.

Accordingly, higher ESG Combined Scores represent stronger CSR reputations, that is, firms that not only demonstrate strong ESG practices but also maintain a low profile in terms of public ESG controversies.

3.3.4. Moderating Variable: Institutional Corruption

Institutional corruption was measured at the country–year level using indicators from the Quality of Government (QoG) Standard Dataset (University of Gothenburg). Our primary measure was the Control of Corruption index from the World Bank’s Worldwide Governance Indicators (WGI), which captures perceptions of public sector corruption—ranging from petty bribery to elite “state capture”. This measure is widely used in studies of institutional integrity and governance quality and is constructed using aggregated expert and survey-based assessments from multiple data sources [44]. WGI scores range from −2.5 (weak control) to +2.5 (strong control). For analytical consistency with our hypotheses, we reverse-coded this variable so that higher values would reflect higher perceived corruption.

To strengthen our robustness checks, we also drew on the Bayesian Corruption Indicators (BCI), developed by Standaert (2015) [45]. The BCI combines more than 100 corruption-related survey items using a Bayesian state-space modeling framework to generate an absolute scale ranging from 0 to 100, where higher values indicate more severe perceived corruption. Unlike the WGI, which provides a relative score, the BCI offers statistically comparable estimates over time and across countries, enabling finer-grained institutional diagnostics. Its methodological rigor and broader input base make it a valuable complement to WGI-based measures, particularly in longitudinal or multi-country research settings.

3.4. Control Variables

To isolate the effects of our key variables of interest, we included a set of control variables to account for firm-level and market-level heterogeneity likely to influence market share outcomes.

At the firm level, we controlled for firm size, measured as the natural logarithm of total revenue, to account for scale-related advantages that may enhance competitiveness. We also included R&D intensity, operationalized as the ratio of R&D expenditure to total sales, capturing the firm’s innovation capacity—a factor that may directly affect product quality and differentiation. In addition, advertising intensity—measured as advertising expenditure relative to sales—was included to control for marketing investments that can influence consumer awareness and brand preference. We also introduced make fixed effects to control for the unobserved, time-invariant, brand-specific attributes, such as reputation or long-standing product positioning, that may systematically affect market share.

At the country market level, we controlled for market size, calculated as the natural logarithm of total vehicle sales in each country–year, to account for variations in the baseline market demand. We also controlled for country-level GDP, measured as the natural logarithm of each country’s gross domestic product, to capture economic conditions that may affect overall purchasing power and vehicle consumption. To account for demographic influences, we included population density, defined as the number of people per square kilometer, and population growth, measured as the annual percentage change in total population. Both of these variables were drawn from the World Development Indicators database and help capture differences in consumer behavior and growth potential across national markets.

Additionally, we included competitor recall frequency, which captures the recall activity of other automobile manufacturers in the same country and year. This variable was measured as the natural logarithm of the total number of recalls issued by all competing makes within the same market–year; it serves to control for broader industry-level reputational spillovers that could influence consumers’ general trust in the safety and quality of automobiles.

Finally, we include year fixed effects to account for global temporal shocks—such as macroeconomic fluctuations, fuel price volatility, or regulatory changes—that may influence all firms in a given year.

Table 1. Variable measurement and data sources.

Type	Variable Name	Measurement Description and Data Sources
Dependent	Market Share	Percentage (0–100 scale) of total new vehicle sales in a given country–year accounted for by each make Data sources: Ward’s AutoWorld
Independent	Recall Intensity	Natural logarithm of one plus the number of recalls issued for a given make in a given year Data sources: Official national regulatory agencies
Moderator	CSR Reputation	ESG Combined Score (lagged, firm-level); incorporates both ESG performance and controversies Data sources: Refinitiv ESG
Moderator	Institutional Corruption	Perceived corruption at the country–year level; measured using WGI and BCI (0–100 scale) Data sources: Quality of Government (QoG)
Control	Firm Size	Natural logarithm of total annual firm revenue (USD) Data sources: Compustat Global
Control	R&D Intensity	Ratio of R&D expenditure to total sales Data sources: Compustat Global
Control	Advertising Intensity	Ratio of advertising expenditure to total sales Data sources: Compustat Global
Control	Market Size	Natural logarithm of total vehicle sales in each country–year Data sources: Ward’s AutoWorld
Control	Country GDP	Natural logarithm of annual gross domestic product (GDP) in each country Data sources: Quality of Government (QoG)
Control	Population Density	Number of people per square kilometer of land area Data sources: Quality of Government (QoG)
Control	Population Growth	Annual percentage change in total population Data sources: Quality of Government (QoG)
Control	Competitor Recall Frequency	Natural logarithm of the total number of recalls issued by competing makes in each market–year Data sources: Official national regulatory agencies

provides a summary of all variables used in the analysis, including their names and measurement descriptions.

3.5. Model Specification and Data Analysis

Given the nested panel structure of our dataset—with automotive makes nested within companies and countries (markets) over multiple years—we employed a series of panel regression techniques to test our hypotheses. Our primary specification used random-effects models, which allowed for the estimation of effects from both time-varying and time-invariant predictors, while accounting for unobserved heterogeneity across makes. To address potential endogeneity concerns and to control for time-invariant unobservables at the make level, we also estimated fixed-effects models as robustness checks. In addition, we implemented mixed-effects models with random intercepts at the country level to account for the hierarchical clustering of makes within countries and to control for unobserved country-specific institutional factors.

All models used robust standard errors clustered at the make level to address potential issues of heteroskedasticity and serial correlation. The sample was restricted to observations through to the year 2015 to ensure the availability of subsequent market share data for all make–country combinations.

The dependent variable in all models was the natural logarithm of market share in year t + 1, measured at the make–country level. Log transformation corrected for the right-skewed distribution of market share and enabled the interpretation of coefficients in terms of proportional change. To account for performance persistence and baseline brand strength, we included lagged market share (in level form) as a predictor. This modeling strategy allowed us to assess how product recalls and CSR reputation affect changes in market share net of prior performance dynamics.

Our specification incorporated several methodological features to enhance model validity. First, the inclusion of lagged market share addressed dynamic effects and mean reversion tendencies in market share evolution. While this approach with random effects could theoretically introduce bias in short panels, our relatively long time series (T = 9) mitigated this concern substantially. Second, our use of make fixed effects controlled for time-invariant brand characteristics, while year fixed effects captured common temporal shocks affecting all markets simultaneously. Third, the three-level structure (firm–country–year) appropriately reflected the reality that multinational automakers operate multiple brands across diverse national markets, allowing us to leverage both within-brand variation over time and cross-country institutional differences.

To test Hypothesis 1, we included an interaction term between recall intensity and CSR reputation. A positive and statistically significant coefficient on this interaction would indicate that firms with stronger CSR reputations are better insulated from the market penalties typically associated with product recalls—consistent with the “insurance-like” function of CSR.

To test Hypothesis 2, we estimated a three-way interaction among recall intensity, CSR reputation, and institutional corruption. Our primary measure of institutional corruption was the Control of Corruption index from the World Bank Worldwide Governance Indicators (WGI), where higher values represent stronger control of corruption (i.e., lower levels of perceived corruption). Under this scale, a negative and statistically significant coefficient on the three-way interaction term would support the hypothesis that CSR’s protective effect is amplified in countries with weaker institutional integrity (i.e., higher corruption).

As a robustness check, we replicated the analysis using the Bayesian Corruption Indicator (BCI), an alternative corruption measure constructed using a Bayesian state-space model that aggregates more than 100 survey-based indicators. In this measure, higher values reflect greater perceived corruption. Thus, a positive and significant three-way interaction using the BCI would offer converging evidence in support of our institutional moderation hypothesis.

4. Results

4.1. Descriptive Statistics

Table 2. Summary Statistics.

		Mean	S.D.	Min	Max
1	Market Share	5.29	9.13	0	76.66
2	Recall Intensity	1.05	0.85	0	4.26
3	CSR Reputation	47.49	13.72	28.39	86.82
4	Institutional Corruption	1.38	0.72	−0.59	2.07
5	Institutional Corruption *	25.79	10.56	16.01	50.18
6	Firm Size	11.44	0.72	9.45	12.52
7	R&D Intensity	7.8	1.86	0	9.55
8	Advertising Intensity	0.65	1.15	0	3.54
9	Market Size	15.06	0.95	13.75	16.97
10	Country GDP	28.84	0.87	27.47	30.49
11	Population Density	183.63	165.1	2.69	520.59
12	Population Growth	0.64	0.63	−1.85	2.06
13	Competitor Recall Frequency	4.58	0.8	1.1	6

* Alternative measure (BCI).

presents the descriptive statistics for the variables used in our analysis. Market share, expressed in percentage points, has a mean of 5.29 (SD = 9.13), with a wide range from 0 to 76.66, reflecting substantial variation in brand-level market presence across country–years. The primary independent variable, recall intensity, measured as the natural logarithm of one plus the number of recalls for each make–year, has a mean of 1.05 (SD = 0.85), indicating considerable variation in recall exposure across brands and time.

CSR reputation, measured using the Refinitiv ESG Combined Score, ranges from 28.39 to 86.82, with a mean of 47.49 (SD = 13.72), reflecting significant heterogeneity in environmental, social, and governance performance among the global automotive firms in our sample.

Our main measure of institutional corruption, based on the World Bank Control of Corruption index, has a mean of 1.38 (SD = 0.72), with lower values indicating greater perceived corruption. The alternative measure, the Bayesian Corruption Indicator (BCI), ranges from 16.01 to 50.18 and has a mean of 25.79 (SD = 10.56), confirming notable cross-country variation in the corruption levels.

Among the firm-level controls, firm size (log revenue) averages 11.44 (SD = 0.72), R&D intensity averages 7.8 (SD = 1.86), and advertising intensity averages 0.65 (SD = 1.15).

At the market level, market size (log national vehicle sales in each country–market) has a mean of 15.06 (SD = 0.95), and Country GDP (log GDP) averages 28.84 (SD = 0.87). Population density varies widely, with a mean of 183.63 (SD = 165.1) and values ranging from 2.69 to 520.59 people per square kilometer. Population growth has a mean of 0.64% (SD = 0.63%), ranging from −1.85% to 2.06%.

Finally, competitor recall frequency, measured as the log of total recalls by rival makes in the same market–year, has a mean of 4.58 (SD = 0.80), suggesting variance in industry-wide recall exposure across markets and years.

Table 3. Correlation Matrix.

	1	2	3	4	5	6
1	1
2	0.18	1
3	−0.16	−0.07	1
4	0.07	0.22	−0.03	1
5	−0.05	−0.17	0.02	−0.94	1
6	0.25	0.01	−0.27	0	0.01	1
7	0.04	0.12	−0.24	0.02	−0.01	0.66
8	0.14	0.22	−0.05	0.01	0	0.1
9	−0.03	0.04	0	−0.53	0.5	0.01
10	−0.01	0.11	0	−0.27	0.28	0.02
11	0.01	−0.12	0.02	−0.32	0.25	0.03
12	−0.03	−0.01	0	0.18	−0.07	−0.01
13	0.02	0.27	−0.09	−0.01	0.01	0.17
	7	8	9	10	11	12	13
7	1
8	0.19	1
9	0.01	0.03	1
10	0.02	0.04	0.94	1
11	0.01	−0.02	−0.09	−0.15	1
12	0	−0.01	−0.35	−0.34	−0.56	1
13	0.45	0.36	0.03	0.02	−0.01	0.01	1

Correlation coefficients greater than 0.2 are statistically significant at the 5% level.

presents the correlation matrix for our key variables. The correlations among our independent variables are generally low to moderate, suggesting minimal concerns about multicollinearity. As expected, the control of corruption measure is strongly negatively correlated with the alternative corruption measure (r = −0.94), confirming that these measures capture similar but oppositely coded constructs. Firm size shows a positive correlation with market share (r = 0.25) and with R&D intensity (r = 0.66), suggesting that larger firms tend to have a greater market presence and invest more in innovation. Interestingly, there is a negative correlation between CSR reputation and firm size (r = −0.27), which might reflect that some smaller manufacturers pursue stronger CSR strategies.

4.2. Regression Results

To test our hypotheses concerning the insurance-like effect of CSR during product recalls and the moderating role of institutional corruption, we estimated panel regression models with robust standard errors. These models account for unobserved heterogeneity across makes and time while addressing potential heteroskedasticity.

Table 4. Regression results (random-effect models).

	Model 1	Model 2	Model 3	Model 4
Market Share	−0.232 ***	−0.151 ***	−0.151 ***	−0.152 ***
	(0.00351)	(0.0195)	(0.0193)	(0.0187)
Recall Intensity		−0.533 **	−2.076 **	−7.469 ***
		(0.263)	(0.842)	(2.316)
CSR Reputation	0.0167 **	0.0214	−0.0121	−0.0548
	(0.00836)	(0.0196)	(0.0297)	(0.0726)
Recall Intensity × CSR Reputation			0.0341 **	0.136 ***
			(0.0148)	(0.0416)
Institutional Corruption (WGI)	−0.807 ***	−1.487 **	−1.592 ***	−3.162
	(0.192)	(0.581)	(0.574)	(2.699)
Recall Intensity × Institutional Corruption (WGI)				3.586 ***
				(1.366)
CSR Reputation × Institutional Corruption (WGI)				0.0291
				(0.0448)
Recall Intensity × CSR Reputation × Institutional Corruption (WGI)				−0.0679 ***
				(0.0257)
Firm Size	1.604 **	0.534	0.422	0.394
	(0.730)	(1.117)	(1.016)	(1.053)
R&D Intensity	−0.0524	0.0574	0.0179	0.00172
	(0.0849)	(0.304)	(0.318)	(0.320)
Advertising Intensity	0.445 **	0.723 **	0.730 **	0.696 **
	(0.183)	(0.314)	(0.326)	(0.325)
Market Size	0.377 *	−1.107	−1.302	−1.342
	(0.207)	(1.126)	(1.132)	(1.102)
Country GDP	−0.478 ***	0.279	0.443	0.572
	(0.148)	(0.889)	(0.897)	(0.851)
Population Density	0.000553	−0.00259 *	−0.00284 *	−0.00237
	(0.000685)	(0.00151)	(0.00147)	(0.00146)
Population Growth	0.325 ***	−0.0130	−0.0613	0.0450
	(0.0910)	(0.264)	(0.267)	(0.277)
Competitor Recall Frequency	0.707	1.293	1.183	1.315
	(0.471)	(1.201)	(1.178)	(1.194)
Year fixed effects	Included	Included	Included	Included
Make fixed effects	Included	Included	Included	Included
Constant	−2.621	7.502	9.537	8.369
	(10.78)	(13.36)	(12.85)	(11.79)
Observations	1517	1192	1192	1192

Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the results using random-effects specifications, with market share (rescaled to percentage points) as the dependent variable.

Model 1 serves as the baseline specification, including only control variables and lagged market share. The coefficient for lagged market share is negative and statistically significant (β = −0.232, p < 0.001), suggesting a regression-to-the-mean effect; makes with exceptionally high or low market shares in one period tend to experience partial reversion in the subsequent period.

Model 2 introduces the main independent variable, recall intensity, measured as the natural logarithm of one plus the number of recalls. The coefficient is negative and significant (β = −0.533, p < 0.05), with a standardized effect size of approximately 0.15 standard deviations, indicating a small to medium effect. This confirms that recalls, on average, lead to subsequent declines in the market share.

Model 3 tests Hypothesis 1, which posits that the negative impact of recalls is mitigated by higher CSR reputation. The interaction term between recall intensity and CSR reputation is positive and statistically significant (β = 0.0341, p < 0.05), supporting the hypothesized insurance-like effect of CSR. To interpret this interaction, we conducted a marginal effects analysis examining how recall impacts vary across different levels of CSR reputation. We calculated the conditional effect of recall intensity on future market share at one standard deviation below and above the mean of CSR reputation. Firms with low CSR reputations (1 SD below mean = 33.77) experience a market share decline of 2.85 percentage points per recall incident. In contrast, firms with high CSR reputations (1 SD above mean = 61.21) experience only a 0.82 percentage point decline. This difference of 2.03 percentage points represents a 71% reduction in recall-related market share loss for high-CSR firms relative to low-CSR firms, demonstrating a substantial buffering effect with both statistical and practical significance.

Model 4 tests Hypothesis 2 by including the following three-way interaction term: Recall Intensity × CSR Reputation × Institutional Corruption (WGI). This interaction term is negative and statistically significant (β = −0.0679, p < 0.01). Because higher values on the WGI index reflect a stronger control of corruption (i.e., lower corruption), the negative coefficient indicates that CSR’s protective effect is stronger in more corrupt environments—consistent with Hypothesis 2. To interpret this three-way interaction and quantify its effect size, we calculated the marginal effects of recall intensity on market share for firms with high CSR reputation (1 SD above mean) across contrasting institutional contexts. In high-corruption environments (1 SD below WGI mean), the marginal effect of recalls is −0.3 percentage points. In low-corruption environments (1 SD above WGI mean), the marginal effect increases to −1.5 percentage points.

This difference of 1.2 percentage points represents a large effect size (Cohen’s d ≈ 0.8). The fivefold difference in recall impact (−0.3 vs. −1.5) demonstrates that CSR’s insurance value is substantially amplified in corrupt environments. Specifically, high-CSR firms in high-corruption contexts experience 80% less market share damage from recalls compared to their counterparts in low-corruption contexts. This finding underscores that CSR’s protective function is highly context-dependent; in environments where institutional trust is lacking and public-sector integrity is low, firms derive significantly greater reputational value from their CSR investments.

Overall, these findings suggest that CSR serves not only as a reputational buffer in general but also as a context-sensitive strategic resource that becomes increasingly valuable in institutional environments where formal governance mechanisms are weak, and stakeholders rely more heavily on firm-level signals of trustworthiness.

The control variables behave largely as expected. Advertising intensity has a consistently positive and statistically significant effect across models (β ≈ 0.7, p < 0.05), underscoring the role of marketing in defending or growing market share. Population density shows a weakly negative effect in some models (e.g., Model 2, p < 0.10), possibly due to intensified competition in denser urban markets. Most other control variables, including R&D intensity, market size, and country GDP, do not exhibit significant effects in the full models. Year and make fixed effects are included in all specifications.

Table 5. Regression results (fixed effect models).

	Model 5	Model 6	Model 7
Market Share	−0.151 ***	−0.151 ***	−0.152 ***
	(0.0194)	(0.0192)	(0.0186)
Recall Intensity	−0.533 *	−2.076 **	−7.469 **
	(0.262)	(0.839)	(2.308)
CSR Reputation	0.0214	−0.0121	−0.0548
	(0.0195)	(0.0296)	(0.0724)
Recall Intensity × CSR Reputation		0.0341 **	0.136 **
		(0.0147)	(0.0414)
Institutional Corruption (WGI)	−1.487 **	−1.592 **	−3.162
	(0.579)	(0.572)	(2.690)
Recall Intensity × Institutional Corruption (WGI)			3.586 **
			(1.361)
CSR Reputation × Institutional Corruption (WGI)			0.0291
			(0.0447)
Recall Intensity × CSR Reputation × Institutional Corruption (WGI)			−0.0679 **
			(0.0256)
Firm Size	0.534	0.422	0.394
	(1.113)	(1.013)	(1.049)
R&D Intensity	0.0574	0.0179	0.00172
	(0.303)	(0.317)	(0.319)
Advertising Intensity	0.723 **	0.730 *	0.696 *
	(0.313)	(0.325)	(0.324)
Market Size	−1.107	−1.302	−1.342
	(1.122)	(1.128)	(1.098)
Country GDP	0.279	0.443	0.572
	(0.886)	(0.894)	(0.848)
Population Density	−0.00259	−0.00284 *	−0.00237
	(0.00151)	(0.00146)	(0.00145)
Population Growth	−0.0130	−0.0613	0.0450
	(0.263)	(0.266)	(0.276)
Competitor Recall Frequency	1.293	1.183	1.315
	(1.196)	(1.174)	(1.190)
Year fixed effects	Included	Included	Included
Make fixed effects	Included	Included	Included
Constant	7.651	9.776	8.559
	(13.48)	(12.97)	(11.89)
Observations	1192	1192	1192
Log likelihood	−3885	−3884	−3880
R-squared	0.250	0.253	0.257

Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the results of fixed-effects regression models, again using market share in the following year as the dependent variable. Models 5–7 replicate the structure of the random-effects models and provide a more conservative test of our hypotheses by controlling for time-invariant, brand-specific heterogeneity.

The pattern of results in the fixed-effects models mirrors those observed in the random-effects models reported in Table 4. Most notably, the three-way interaction between recall intensity, CSR reputation, and institutional corruption remains statistically significant (Model 7: β = −0.0679, p < 0.05), providing continued support for Hypothesis 2—that the reputational buffering effect of CSR is amplified in more corrupt institutional environments.

In Model 6, the two-way interaction between recall intensity and CSR reputation is again positive and statistically significant (β = 0.0341, p < 0.05), consistent with Hypothesis 1. The interaction between recall intensity and institutional corruption is also significant and positive (β = 3.586, p < 0.05), reinforcing the notion that the institutional environment shapes how stakeholders perceive and respond to product recalls. These two-way interactions suggest that both CSR and institutional quality independently moderate the impact of recalls on market outcomes.

In Model 7, we introduced the full three-way interaction among recall intensity, CSR reputation, and institutional corruption to test Hypothesis 2. The coefficient on the three-way interaction term is negative and statistically significant (β = −0.0679, p < 0.05). Given that the Control of Corruption index was coded such that higher values would reflect stronger governance (i.e., lower corruption), the negative sign on this coefficient implies that the insurance-like effect of CSR is amplified in more corrupt institutional environments. In other words, the reputational protection offered by CSR becomes more pronounced where formal institutions are weaker and stakeholders are likely to place greater reliance on firm-level signals of integrity and responsibility.

Together, these findings provide strong support for both Hypotheses 1 and 2. They indicate that while CSR generally buffers firms from the market penalties of recall events, this effect is not uniform across institutional contexts. Instead, the protective value of CSR is contingent on the level of institutional corruption, with the greatest benefits accruing to firms operating in low-trust, high-corruption environments.

Table 6. Robustness checks (alternative measure of corruption, mixed effects models).

	Model 8	Model 9	Model 10
Market Share	−0.153 ***	−0.153 ***	−0.161 **
	(0.0186)	(0.0185)	(0.0690)
Recall Intensity	4.511 *	4.511	−7.757 **
	(2.535)	(2.526)	(3.891)
CSR Reputation	0.0220	0.0220	−0.0621
	(0.0713)	(0.0711)	(0.0397)
Recall Intensity × CSR Reputation	−0.0836 *	−0.0836	0.143 **
	(0.0466)	(0.0464)	(0.0677)
Institutional Corruption (WGI)			−3.269 ***
			(1.094)
Recall Intensity × Institutional Corruption (WGI)			3.948 **
			(1.932)
CSR Reputation × Institutional Corruption (WGI)			0.0337 *
			(0.0181)
Recall Intensity × CSR Reputation × Institutional Corruption (WGI)			−0.0734 **
			(0.0368)
Institutional Corruption (BCI)	0.136	0.136
	(0.175)	(0.175)
Recall Intensity × Institutional Corruption (BCI)	−0.268 ***	−0.268 **
	(0.0999)	(0.0995)
CSR Reputation × Institutional Corruption (BCI)	−0.00130	−0.00130
	(0.00291)	(0.00290)
Recall Intensity × CSR Reputation × Institutional Corruption (BCI)	0.00472 ***	0.00472 **
	(0.00182)	(0.00181)
Firm Size	0.399	0.399	0.553
	(1.052)	(1.048)	(0.644)
R&D Intensity	0.00146	0.00146	0.0474
	(0.317)	(0.316)	(0.174)
Advertising Intensity	0.729 **	0.729 *	0.626 *
	(0.327)	(0.325)	(0.333)
Market Size	0.299	0.299	−0.984 ***
	(0.825)	(0.822)	(0.285)
Country GDP	−0.720	−0.720	0.218
	(0.592)	(0.590)	(0.532)
Population Density	−0.000677	−0.000677	−0.00246
	(0.00126)	(0.00126)	(0.00320)
Population Growth	0.366	0.366	−0.0986
	(0.332)	(0.330)	(0.242)
Competitor Recall Frequency	1.373	1.373	1.150
	(1.149)	(1.145)	(1.245)
Make fixed effects	Included	Included	Included
Year fixed effects	Included	Included	Included
Constant	11.89	12.29	12.11
	(12.87)	(12.95)	(10.17)
Observations	1192	1192	1192
Log likelihood		−3879	−3874
R-squared		0.258

Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

presents additional robustness checks using alternative modeling strategies and a different corruption measure—the Bayesian Corruption Indicator (BCI). In this index, higher values reflect greater perceived corruption, contrasting with the original WGI-based measure.

Models 8 and 9 replicate the random- and fixed-effects specifications, respectively, using BCI as the moderator. In both models, the three-way interaction between recall intensity, CSR reputation, and institutional corruption (BCI) is positive and statistically significant (Model 8: β = 0.00472, p < 0.001; Model 9: β = 0.00472, p < 0.05), supporting Hypothesis 2 from the opposite coding direction. These results confirm that the insurance-like effect of CSR is most pronounced in high-corruption environments, regardless of which corruption measure, is used.

Model 10 introduces a mixed-effects specification with country-level random intercepts, accounting for the nested structure of make–level observations within countries. Here too, the three-way interaction (β = −0.0734, p < 0.05) remains statistically significant, further reinforcing the robustness of our findings.

Across Models 6 through 10, several control variables exhibit significant and consistent effects. Advertising intensity is positively and significantly associated with future market share (β ≈ 0.7, p < 0.1 or better), highlighting the role of marketing expenditures in shaping competitive outcomes. Population density shows a weakly negative relationship with market share in some models, possibly due to intensified brand competition in urbanized or saturated markets.

Finally, the inclusion of make fixed effects captures unobserved heterogeneity across brands, most of which remain statistically significant, indicating persistent brand-level differences not fully explained by the included predictors.

5. Discussion and Conclusions

5.1. Discussion

This study examines how corporate social responsibility (CSR) shapes firms’ resilience during product recalls and how this effect varies across institutional contexts. Focusing on the global automotive industry, we find that CSR reputation provides a robust insurance-like effect—significantly mitigating the negative market share consequences of recalls. Notably, this protective effect is not uniform; it is substantially stronger in countries with higher levels of institutional corruption. This suggests that CSR becomes a particularly valuable organizational asset in environments where formal institutions are weak and stakeholder skepticism toward both corporate and governmental actors is elevated.

5.2. The Insurance-like Effect of CSR

Our first key finding confirms that CSR reputation moderates the relationship between product recalls and subsequent market performance. Firms with higher CSR ratings—as captured by ESG scores—experience significantly smaller declines in market share following recalls. This provides empirical support for the “insurance-like” function of CSR, originally theorized by Godfrey et al. (2009) and extended by Minor and Morgan (2011) [6,7]. Quantitatively, firms with ESG scores one standard deviation above the sample mean experienced approximately 38% to 71% less market share loss compared to those one standard deviation below, depending on the model specification. This finding reinforces the idea that CSR is not merely symbolic but yields tangible strategic benefits during reputational crises.

The buffering effect of CSR can be explained through three interrelated mechanisms. First, consistent with moral capital theory, CSR activities generate stakeholder goodwill that firms can draw upon in times of adversity. This accumulated trust functions as a reservoir of moral capital, shielding high-CSR firms from the full reputational fallout of recall events. Second, attribution theory helps explain how stakeholders interpret corporate actions through the lens of prior reputation. When firms with strong CSR records initiate recalls, these actions are more likely to be seen as expressions of ethical responsibility than as reactive maneuvers to avoid liability. This favorable attribution helps sustain consumer confidence. Third, CSR acts as a credible signal of long-term stakeholder commitment. Firms that consistently invest in CSR demonstrate a prioritization of safety and public welfare over short-term profits, making their recall announcements more likely to be perceived as authentic and proactive.

5.3. The Moderating Role of Institutional Corruption

Our most theoretically novel finding is that the insurance-like effect of CSR is significantly amplified in environments characterized by high institutional corruption. The three-way interaction between recall intensity, CSR reputation, and institutional corruption reveals that the protective value of CSR is contingent on the credibility of formal institutions. In high-corruption contexts, firms with strong CSR performance saw the negative effect of recalls reduced by approximately 70%, whereas in low-corruption contexts, the reduction was only around 30%. This suggests that when stakeholders cannot rely on regulatory systems to ensure accountability, they place greater weight on firm-level signals of integrity.

This finding contributes to the literature on the contextual value of CSR by demonstrating that institutional weaknesses, while often viewed as obstacles to CSR implementation, can in fact enhance the strategic benefits of CSR investments. In corrupt environments, CSR not only compensates for institutional voids but also becomes a critical differentiator that shapes stakeholder attribution and market response. Moreover, our study advances attribution theory by showing that stakeholder inferences are shaped not only by firm-level cues but also by the institutional environment in which firms operate. In low-trust settings, where unethical behavior is presumed to be widespread, CSR provides a distinctive and credible basis for positive stakeholder attribution during crises.

5.4. Theoretical Implications

Our findings contribute to multiple theoretical domains while extending and refining the existing literature.

First, we refine the insurance-like view of CSR by revealing critical boundary conditions for its buffering effect. While prior research established that CSR generates moral capital, thereby providing insurance-like protection during crises, and subsequent studies confirmed this effect [7,16], our research demonstrates that this protection is not universally constant. Instead, it varies systematically with institutional corruption levels, being significantly amplified in high-corruption environments. This finding challenges the implicit assumption in prior work that CSR’s protective value operates uniformly across contexts. Our results suggest that firms in corrupt environments derive up to five times greater reputational benefit from CSR during recalls compared to those in low-corruption settings, calling for a more nuanced, context-sensitive understanding of when and where CSR investments yield maximum strategic returns.

Second, we extend institutional theory by demonstrating how firm-level CSR capabilities serve as a substitute for weak formal institutions. Building on Rodriguez et al.’s (2006) framework of CSR in the non-market environment, as well as Khanna and Palepu’s work on institutional voids [8,23], we show that CSR serves as an alternative trust-building mechanism when public governance fails. Our findings align with Matten and Moon’s (2008) implicit/explicit CSR framework [6], suggesting that in contexts where formal institutions underperform, CSR takes on heightened strategic importance as firms fill governance gaps. This substitution effect is particularly pronounced during crises; when institutional corruption is high, stakeholders rely more heavily on firm-specific signals like CSR reputation to assess trustworthiness, as formal regulatory oversight becomes less credible.

Third, our research advances crisis management and attribution theory by identifying institutional corruption as a critical contextual moderator. While Coombs (2007) emphasized how prior reputation shapes crisis responses [35], and Klein and Dawar (2004) demonstrated that CSR influences stakeholder attributions during product-harm events [17], we show that these effects are fundamentally shaped by the institutional environment. In high-corruption contexts, stakeholders appear more willing to attribute recalls to external factors rather than managerial failure when firms have strong CSR reputations. This extends attribution theory by revealing how macro-level institutional factors interact with firm-level reputation to shape stakeholder judgments during crises.

Fourth, our findings contribute to the emerging market literature by challenging the conventional knowledge about CSR’s applicability. Contrary to views that CSR is primarily a developed-market phenomenon, our results echo the suggestion of Jamali and Mirshak’s (2007) that CSR may be even more strategically valuable in institutionally weak environments [26]. The finding that CSR’s insurance effect is amplified in corrupt settings aligns with El Ghoul et al.’s (2017) evidence that CSR creates greater financial value in countries with weaker market institutions [25]. This suggests that rather than being a “luxury” for firms in advanced economies, CSR may represent a critical strategic investment for navigating the institutional challenges characteristic of emerging markets.

Finally, our study offers practical implications for crisis management in global markets. The strong interactions among CSR reputation, recalls, and institutional corruption suggest that multinational firms should calibrate their CSR investments and crisis response strategies to local institutional contexts. In high-corruption environments where our results show CSR provides maximum protective value, firms may find greater returns from CSR investments than in well-governed markets where formal institutions already provide stakeholder protection.

5.5. Practical Implications

This study yields several important practical implications for managers and policymakers navigating complex global markets.

First, CSR should be reconceptualized as strategic risk insurance rather than mere philanthropy. Our findings demonstrate that firms with strong CSR reputations experience up to 71% less market share decline following recalls compared to low-CSR firms. This substantial buffering effect suggests that CSR investments function as a form of reputational insurance premium, generating measurable returns during crisis events. Managers should therefore evaluate CSR expenditures not only through traditional cost–benefit analyses but also as strategic investments in crisis resilience that protect long-term market position.

Second, multinational corporations should fundamentally reconsider their global CSR allocation strategies. Our results reveal that CSR delivers approximately five times greater protective value in high-corruption environments compared to well-governed markets. This challenges the conventional practice of concentrating CSR investments in developed markets where stakeholder pressure is highest. Instead, firms should consider prioritizing CSR initiatives in institutionally weak markets where the strategic returns are demonstrably higher. For instance, a manufacturer facing budget constraints might achieve greater risk mitigation by enhancing CSR activities in emerging markets rather than expanding programs in established Western markets.

Third, managers must resist the temptation to cut CSR budgets during economic downturns or financial pressures. Our findings show that CSR’s protective value becomes most critical precisely during crisis periods when firms are most vulnerable to reputational damage. The insurance-like benefits we document—which can preserve billions in market value during major recalls—are often invisible in quarterly financial statements but become starkly apparent during adverse events. CFOs and boards should therefore treat CSR expenditures as essential investments in organizational resilience rather than discretionary expenses vulnerable to cost-cutting measures.

Fourth, crisis communication strategies must be tailored to institutional contexts, with CSR credentials playing a differentiated role. In high-corruption environments where institutional trust is low, our results suggest that emphasizing a firm’s CSR track record during recall announcements can be particularly effective in maintaining stakeholder confidence. Firms should develop market-specific crisis playbooks that leverage CSR reputation more prominently in corrupt markets while relying more on regulatory compliance messaging in well-governed contexts. For example, recall communications in markets with weak institutions should lead with the firm’s demonstrated commitment to social responsibility, while those in strong institutional environments might emphasize regulatory cooperation and technical remediation.

Finally, policymakers in high-corruption countries should recognize CSR as a complementary governance mechanism. Rather than viewing private CSR initiatives with suspicion, governments in institutionally weak environments might achieve better market outcomes by incentivizing genuine CSR engagement through tax benefits or regulatory flexibility. This could create a virtuous cycle where firms are rewarded for filling institutional voids, ultimately benefiting both market stability and consumer welfare.

5.6. Limitations and Future Research

Several limitations of this study point to directions for future research. First, while we capture performance outcomes through market share changes, we do not directly measure changes in consumer perceptions. Future work could incorporate brand trust, social media sentiment, or survey data to better capture the cognitive and emotional mechanisms behind CSR’s buffering effect. Second, while the automotive industry provides a compelling empirical setting, further studies could investigate whether similar dynamics are held in other sectors with different product risks and stakeholder configurations, such as pharmaceuticals, food, or technology. Third, although corruption serves as a meaningful institutional moderator, future research could explore how other institutional characteristics, such as regulatory enforcement, media freedom, or judicial independence, moderate the effectiveness of CSR in crisis contexts. Fourth, we focus primarily on short-term outcomes. Longitudinal research could examine how CSR affects firms’ post-recall recovery trajectories over multiple years, including investor behavior, product evaluations, and reputation rebuilding. Fifth, our analysis draws on data from 2007 to 2015, which may not fully capture recent developments in CSR practices or shifts in institutional environments. Nonetheless, this period offers a meaningful context for testing our hypotheses, as it includes a wide range of product recall events across countries with varying levels of institutional corruption. Future research can benefit by revisiting these dynamics using more recent data to examine whether the patterns observed in this study persist under current CSR expectations and regulatory pressures.

5.7. Conclusions

This study advances our understanding of CSR as a strategic asset, whose value is contextually constructed and crisis tested. By analyzing panel data on global automakers across eight countries from 2007 to 2015, we provide robust evidence that CSR significantly reduces the performance penalties associated with product recalls. Crucially, we show that this insurance-like protection is amplified in institutional environments characterized by high levels of corruption.

These findings challenge the notion that CSR yields uniform returns across settings and highlight the importance of considering institutional context when evaluating the strategic value of CSR. As firms navigate an increasingly complex global landscape marked by regulatory uncertainty, stakeholder scrutiny, and recurring reputational risks, understanding how CSR and governance conditions interact becomes ever more important. Our study offers a foundation for more nuanced, context-sensitive approaches to both CSR investment and crisis management strategy.