Informal Institutions and Corporate Reputational Exposure: The Role of Public Environmental Perceptions

Chrysovalantis Gaganis, Panagiota Papadimitri, Fotios Pasiouras and Alexia Ventouri 4 Department of Economics, University of Crete, Rethymno, 74100, Greece, Portsmouth Business School, University of Portsmouth, Portland Street, Portsmouth, PO1 3AH, UK, Department of Financial Management, Law & Accounting, Montpellier Business School, 2300 Avenue des Moulins, Cedex 4, Montpellier, 34185, France, and King’s Business School, King’s College London, 30 Aldwych, London, WC2B 4BG, UK Corresponding author email: alexia.ventouri@kcl.ac.uk


Introduction
In recent years, policy-makers have introduced various initiatives to address environmental concerns and to combat climate change. These often take the form of pressures from a formal institutional environment, like environmental courts and regulations, which can in turn be associated with higher costs and lower profits for firms (Berkman, Jona and Soderstrom, 2019;Zhang, Yu and Kong, 2019), putting pressure on managers. 1 For example, Mark Carney (2019) highlights that the global transition needed to tackle the climate crisis could result in an abrupt 2 C. Gaganis et al. financial collapse, as firms struggle to operate under extreme shifts in the institutional environment. Additionally, managers face further pressure from informal institutions, like social norms and beliefs that enhance public awareness of how firms' activities can have an impact on the physical environment and climate change. Arguably, therefore, firms now operate within an extreme institutional environment where they must balance the interests of shareholders with those of other stakeholder groups. While some recent studies have examined the impact of formal environment-related institutions on firm outcomes (Shi and Xu, 2018;Zhang, Yu and Kong, 2019), less attention has been given to the role of informal institutions. Accordingly, this study examines whether informal institutions impact public perceptions of environmental issues and how such perceptions influence corporate reputation.
Corporate reputation and the associated risks have received much attention from academics and practitioners, illustrating their relationship to market value (Black, Carnes and Richardson, 2000;Weber Shandwick and KRC Research, 2020), perceived importance for executives (Deloitte, 2014) and association with better firm outcomes (e.g. Cao et al., 2015;Roberts and Dowling, 2002). Nonetheless, evidence suggests that only a small proportion of firms feel capable of managing reputational risk (Deloitte, 2014), which might be explained by poor understanding of its sources and how to measure it (Deloitte, 2016).
Therefore, a question that naturally emerges is what drives reputation and reputational risk. Today, most of our knowledge comes from countryspecific studies (mainly focused on the USA) that analyse firm-specific and industry-specific factors like financial performance, size and popular management techniques (Staw and Epstein, 2000), downsizing (Zyglidopoulos, 2005), board characteristics (Musteen, Dattal and Kemmerer, 2010), social performance and the nature of business activities (Brammer and Pavelin, 2006;Nardella, Brammer and Surdu, 2019). In other words, as discussed by Soleimani, Schneper and Newburry (2014), the reputation literature often implicitly treats the driving factors of firm reputation as fixed and universal across countries, while the impact of national institutions on corporate reputation assessment has received comparatively less attention and is still not fully understood. Gardberg (2006) emphasizes that since cor-porate reputation is a social variable in an open system, varying regulatory, normative and cognitive elements of national institutional environments may affect corporate reputation construction and expectations, and thus its international generalizability.
As little is known about the country-level drivers of reputational assessments (Soleimani, Schneper and Newburry, 2014), this study extends the literature on corporate reputational rankings (e.g. Bermiss, Zajac and King, 2014) by examining how public perceptions of environmental issues affect reputational risk. Our analysis is based on institutional theory, which has been widely adopted to explain how firm behaviour is driven by institutional pressures. Despite the wide adoption of institutional theory in studies of environmental management practices (Berrone et al., 2013;Daddi et al., 2016), corporate social responsibility (Brammer, Jackson and Matten, 2012), sustainable practices (Glover et al., 2014) and climate change strategies (Daddi et al., 2016), there has been little empirical investigation of the institutional drivers of reputation (see e.g. Deephouse, Newburry and Soleimani, 2016;Soleimani, Schneper and Newburry, 2014), and no study of the association between environment-related perceptions of the public and reputational exposure. Building on neo-institutionalism theory and upper echelons theory, we also examine the conditional role of the background characteristics of company directors. Thus, we make two contributions to the literature. First, as we discuss in detail in the next section, the impact of public perceptions on environmental issues and reputational exposure is ambiguous, as it could be either positive or negative. This remains an open question to be answered empirically, and we present the first systematic analysis of this issue in a cross-country setting. This is not only a matter of using a large international sample. Most importantly, it allows us to examine whether countrylevel characteristics, like formal institutions (e.g. rule of law, regulatory quality), moderate this relationship. Second, we examine the conditional role of an array of firm-level corporate governance characteristics, like CEO nationality, board age diversity, board qualifications, board nationality mix and board gender mix. Our findings could be of interest to various groups. First, as we present new empirical evidence, they could be of interest to scholars on corporate reputation, corporate governance and institutional theory. Second, as we discuss in more detail, they could be of interest to firm stakeholders like policy-makers, managers, shareholders and prospective investors. The rest of this paper is structured as follows. The following section discusses the theoretical background and develops the hypotheses. The third section then outlines the data and methodology. The fourth section reports the results. Finally, the last section discusses our findings and concludes.

Institutional theory and reputation
Institutional theory emphasizes how the social and cultural pressures imposed on organizations influence their practices and structures (Scott, 1992). Central to this theory is the idea that many aspects of organizations are driven by the desire to achieve fit with the institutional environment. This fit has been defined as 'the degree of compliance by an organisation with the organisational form of structures, routines, and systems prescribed by institutional norms' (Kondra and Hinings, 1998, p. 750). Thus, institutional theory appears integral to the concept of corporate reputation, which represents the collective evaluation by stakeholders of a firm's goals, values and behaviour compared to those of other firms and to the stakeholders' own expectations (Deephouse, Newburry and Soleimani, 2016;Mishina, Block and Mannor, 2012). Within the same context, Soleimani, Schneper and Newburry (2014) highlight that corporate reputation depends on the extent to which firm behaviour conforms with socially constructed beliefs about which goals and objectives should be pursued. Figure 1 outlines our research framework and hypotheses, which we discuss in more detail below.
The traditional view of institutional theory asserts that institutional isomorphism makes organizations quite similar, through a process that leads them to adopt similar forms and practices (DiMaggio and Powell, 1983), and consequently promotes the success and survival of organizations (Meyer and Rowan, 1977). 2 Within this context, Oliver (1991) highlights that the self-serving advantages of compliance with institutional norms and requirements are illustrated by the claimed association between organizational conformity and 2 DiMaggio and Powell (1983) distinguish between three mechanisms through which institutional isomorphic change occurs. The first is coercive isomorphism that stems from political influence and the problem of legitimacy. This results from both formal and informal (e.g. cultural expectations) pressures on organizations. The second mechanism is mimetic isomorphism resulting from standard responses to uncertainty. In this case, organizations adopt mimetic behaviours and tend to model themselves on similar organizations in their field that they perceive as more legitimate or successful. The third mechanism is normative isomorphism, associated with professionalization. Nonetheless, DiMaggio and Powell (1983) also recognize that this typology is analytic and that these types are not always empirically distinct. 4 C. Gaganis et al. the various rewards discussed in the institutional literature, including increased prestige. Wright and Rwabizambuga (2006) also argue that 'firms are rewarded with enhanced legitimacy and reputation if they develop internal structures "isomorphic" with external institutional pressures ' (p. 90). In general, this traditional view of institutional isomorphism and organizational similarities implies that managerial and other firm-level characteristics do not have an important role. For example, according to Oliver (1991), 'institutional theory illustrates how the exercise of strategic choice may be pre-empted when organisations are unconscious of, blind to, or otherwise take for granted the institutional process to which they adhere' (p. 148). Moreover, 'In the face of very widely shared and taken-for-granted understandings of what constitutes legitimate or rational behaviour, organisations will conform largely because it does not occur to them to do otherwise' (Oliver, 1991, p. 169). Along the same lines, Suchman (1995) argues that external institutions, like culture, construct and interpenetrate the organization in every respect. Therefore, the decisions of the managers are often constructed by the same belief systems that determine the reactions of the audience. This leads to a high degree of convergence between institutionalization and legitimacy.
Based on the above discussion, we expect that societal beliefs about the environment will have a direct impact on reputation. Firms have no option but to be institutionalized and behave in accordance with public expectations. With the consequential narrowing of the gap between broader societal expectations and the effects of corporate practices that can challenge the legitimacy and reputation of individual firms, reputational exposure will decline. Therefore, we formulate the first hypothesis as follows: H1a There is a negative relationship between environmentally friendly societal beliefs and corporate reputational exposure.
However, various neo-institutionalism studies (Walls and Hoffman, 2013) argue that firms may respond heterogeneously when subjected to a homogenous level of institutional pressures (Aharonson and Bort, 2015;Oliver, 1991;Wang, Li and Zhao, 2018). At the same time, societies with more environmentally friendly societal beliefs tend to set higher standards for their companies, and therefore firms face a greater risk of falling short of the expectations. In more detail, organizations must devote resources towards environmental initiatives in a way that simultaneously satisfies their economic objectives (Hoffman, 2001). This might conflict with pressures from shareholders to increase profitability. As Oliver (1991) considers, an organization whose performance and survival only moderately depend upon good public opinion might choose avoidance tactics in response to institutional rules and expectations. Within the same context, Chen et al. (2018) mention that high expenditure and unclear future benefit make some firms reluctant to engage in green innovation, even when faced with strong institutional pressures. They also note that institutional pressures are coercive in nature, driven by the threat of either legal sanction or social sanction, like protests, negative press and diminished reputation and image.
Consequently, it is possible that higher expectations are more likely to be violated by firms, leading to poor institutional fit, public criticism and higher reputational exposure. Thus, under this scenario we would expect to find a positive association between public environmental perceptions and corporate reputational exposure. Hence, we formulate the alternative first hypothesis as follows: H1b There is a positive relationship between environmentally friendly societal beliefs and corporate reputational exposure.
Another issue recently discussed in the literature is that the response to institutional pressures depends upon managerial factors. For instance, Walls and Hoffman (2013) propose that the variance in organizational actions towards environmental sustainability depends primarily on the direction set by the board. Wang, Li and Zhao (2018) find that top management's environmental commitment moderates the relationship between institutional pressures and environmental management practices. Along the same lines, focusing on institutional pressures on corporate climate change strategies, Daddi et al. (2020) conclude that companies with higher managerial sensitivity to climate change are more likely to adopt both mitigation and adaption strategies. González-Benito and González-Benito (2008) highlight the role of managers' ability and willingness to monitor and listen to stakeholders' environmental demands. As they discuss, even if stakeholder demands are clearly specified (which is not always the case), managers might differ in their level of attention and ascribed importance. In some cases, managers might overstate the consequences of ignoring such demands; in other cases, they could underestimate them. These arguments and findings appear to be in line with upper echelons theory, which states that organizational outcomes, including both strategic choices and performance levels, are partially predicted by the background characteristics of the top management team (Hambrick and Mason, 1984). Therefore, we formulate the second hypothesis as follows: H2 Background characteristics of the board of directors moderate the relationship between environmentally friendly societal beliefs and corporate reputational exposure. The dependent variable is Reputational exposure rating, an indicator of ESG-related reputational risk maintained by the business intelligence provider RepRisk. The rating assesses the ESG risk exposure of companies worldwide by systematically capturing negative incidents, criticism and controversies on a daily basis from over 80,000 media outlets, stakeholders and third-party sources in 20 languages; these sources include all major print and online media, non-governmental organizations, regulators, news sites, governmental agencies and social media. 9 RepRisk gathers data through a five-step process: (i) screening; (ii) identification and filtering; (iii) analysis; (iv) quality assurance; and (v) quantification. Once an incident is identified, the analysis includes verification of its type and nature, as well as classification into one or more of the 28 predefined ESG categories. RepRisk argues that its analysis is issuesdriven, rather than firm-driven, and so does not

Dependent variable
Reputational exposure rating Reputational risk rating that reflects: (i) a company's own ESG-related risk exposure due to risk incidents reported specifically about the company, and (ii) the country-sector ESG risk that takes into consideration the sector and location of the company's headquarters and countries where the company has been exposed to ESG risk incidents.
The variable takes a value from 1 to 10, where D = 1 and AAA = 10. Higher values indicate a better rating and lower reputational exposure related to ESG issues.

Independent variables
Environmental perceptions (ECP) Index of public perceptions related to the environment, calculated from the three sub-indices outlined below. It takes a value between 0 and 100, with higher values reflecting more environmentally friendly responses.

ESS Survey (2016)
Energy Sub-index of public perceptions related to energy, based on the answers of individuals in each country to the following questions: (i) 'How likely are you to buy the most energy-efficient home appliance?'; (ii) 'How often do you do things to reduce energy use?'; (iii) 'How confident are you that you could use less energy than now?' Answers given on a predetermined scale were first weighted by the corresponding percentage for each response in each country and then standardized to take a value between 0 and 100. Higher values indicate more environmentally friendly responses.

Firm-specific variables
Profitability Net interest before tax divided by total assets.

Datastream
Size Natural logarithm of sales.

Capital
Equity divided by total assets.

Datastream
Board age diversity Standard deviation of the ages of directors.

Boardex
Board qualifications Standard deviation of the total number of qualifications of directors.

Boardex
Board nationality mix Proportion of directors from different countries.

Boardex
Board gender ratio Proportion of male directors.

Boardex
Board size Natural logarithm of the total number of board members at the end of the fiscal year.

Datastream
Board independence Percentage of independent board members.

Datastream
Local CEO Dummy variable that takes the value of one when the CEO is not a foreign national, and zero otherwise.
Orbis CSR sustainability index Dummy variable that takes the value of one when the company belongs to a specific sustainability index, and zero otherwise.

Datastream
Sustainability reporting Dummy variable that takes the value of one when the company publishes a separate sustainability report or a section in its annual report on sustainability, and zero otherwise.

Country-specific variables
GDP growth Gross domestic product (GDP) growth.

WB
Economic globalization Foreign direct investment (% of GDP).

WB
Shareholder rights Extent of Shareholder Rights Index. For each component, a score of zero is assigned if the answer is no, and one if yes: (1) whether the sale of 51% of the buyer's assets requires shareholder approval; (2) whether shareholders representing 10% of the buyer's share capital have the right to call for a meeting of shareholders; (3) whether the buyer must obtain its shareholders' approval every time it issues new shares; (4) whether shareholders automatically receive pre-emption rights when the buyer issues new shares; (5) whether shareholders elect and dismiss the external auditor; (6) whether changes to the rights of a class of shares are only possible if the holders of the affected shares approve; (7) assuming that the buyer is a limited company, whether the sale of 51% of the buyer's assets requires member approval; (8) assuming that the buyer is a limited company, whether members representing 10% have the right to call for a meeting of members; (9) assuming that the buyer is a limited company, whether all or almost all members must consent to add a new member; (10) assuming that the buyer is a limited company, whether members must first offer their interest to the existing members before they can sell to non-members. The index takes a value between 0 and 10, with higher values indicating higher shareholder rights.
WB RISE A set of indicators to help compare national policy and regulatory frameworks for sustainable energy. RISE classifies countries into a green zone of strong performers in the top third of the 0−100 score range, a yellow zone of middle-third performers and a red zone of weaker performers in the bottom third.

Local competition
Indicator of the intensity of local competition, based on answers to the following question in the Executive Opinion Survey of the World Economic Forum: 'In your country, how intense is competition in the local markets?' This index takes a value between 1 (not intense at all) and 7 (extremely intense).

Buyer sophistication
Indicator of buyer sophistication, based on answers to the following question in the Executive Opinion Survey of the World Economic Forum: 'In your country, on what basis do buyers make purchasing decisions?' This index takes a value between 1 (based solely on the lowest price) and 7 (based on sophisticated performance attributes).

WEF
Labour-employer cooperation Indicator of cooperation in labour-employer relations, based on answers to the following question in the Executive Opinion Survey of the World Economic Forum: 'In your country, how do you characterize labour-employer relations?' This index takes a value between 1 (generally confrontational) and 7 (generally cooperative).

Poverty
Percentage of the population living below national poverty lines.

WEF
Temp Natural logarithm of the difference of monthly temperature over mean, 1979−2010, weighted by land area. Felbermayr and Gröschl (2014) 8 C. Gaganis et al.

Press
Based on responses to the following question: 'Do the major print and broadcast media represent a wide range of political perspectives?' Responses take a value from 0 to 3: 'The major media represent only the government's perspective' scores 0; 'The major media represent only the perspectives of the government and a government-approved, semi-official opposition party' scores 1; 'The major media represent a variety of political perspectives but they systematically ignore at least one political perspective that is important in this society' scores 2; 'All perspectives that are important in this society are represented in at least one of the major media' scores 3.  necessarily focus on a set list of firms, thus assuring some impartiality. The 28 ESG issues drive the entire research process, and every risk incident in RepRisk's ESG risk platform is linked to at least one of these issues. Incident categorizations map to the 10 principles of the United Nations Global Compact, and are related to: (i) environmental footprint (e.g. global pollution, overuse and wasting of resources); (ii) community relations (e.g. human rights abuses); (iii) employee relations (e.g. child labour); (iv) corporate governance (e.g. corruption, executive compensation issues); and (v) crosscutting issues (e.g. controversial products and services). The index also covers 50 ESG 'hot topics', such as palm oil, land mines, deep sea drilling and water scarcity. 10 Each incident is also assigned two proprietary scores based on severity (the magni-10 The 'hot topics' are a dynamic concept, with the list expanding over time in line with developments and client feedback. The list included: Abusive/illegal fishing; Agricultural commodity speculation; Alcohol; Animal transportation; Arctic drilling; Asbestos; Automatic and semi-automatic weapons; Biological weapons; Chemical weapons; Cluster munitions; Coal-fired power plants; Conflict minerals; Coral reefs; Cyberattack; Deep sea drilling; Depleted uranium munitions; Diamonds; Drones; Endangered species; Forest burning; Fracking; tude of the perceived impact) and reach (the influence or readership of source documents). The data only measure negative ESG impacts, and not positive ESG-related events. Although we would ideally consider both positive and negative reported impacts, positive events are less likely to be reported by the media and are mostly self-reported for marketing purposes, making it especially difficult to capture and quantify such data.
The RepRisk rating assigned to each company ranges from AAA (high quality/low risk) to D (low quality/high risk), similar to a credit rating. This rating reflects: (i) a company's own ESGrelated risk exposure due to risk incidents reported specifically about the company, and (ii) the  country-sector ESG risk, which considers the company's sector, the location of its headquarters and any countries where the company has been exposed to ESG risk incidents. Given its ordered nature, we follow the credit rating literature (see e.g. Ashbaugh-Skaife, Collins and LaFond, 2006) by converting the RepRisk ratings into a 10-point numeric scale in which higher numbers indicate higher quality and lower reputational exposure: D/C = 1, C = 2, CC = 3, CCC = 4, B = 5, BB = 6, BBB = 7, A = 8, AA = 9, AAA = 10. Figure 2 shows the national mean values of the dependent variable. The score ranges from 5.52 (Russia) to 9.36 (Czech Republic), with an overall sample mean of 7.65. Thus, there appears to be variation in reputational risk ratings across countries.
Key independent variables. The core variables of interest in our study are the measures of public attitudes on environmental issues. The ESS applies several methodological standards regarding questionnaire design, interview process, translation and data collection. The questionnaire design is developed every 2 years in English, including extensive testing and piloting by national teams (European Social Survey, 2016). Each country needs to achieve a minimum effective sample, representative of the country's population. The national coordinator, the sampling expert and possibly a representative of the survey agency col-lectively devise the optimum sampling design per country. Interviews are conducted face-to-face with individuals aged 15 and over (no upper age limit) residing in private households in each country, regardless of their nationality, citizenship or language. The full questionnaire and the complete ESS Round 8 dataset can be downloaded from http://www.europeansocialsurvey.org.
To construct our variables, we focus on answers to questions that elicit a person's beliefs regarding energy, climate and policy issues. We then use these three sub-indices to create an overall index of Environmental perceptions (also termed ECP). The Energy sub-index is based on answers to the following three questions: (i) 'How likely are you to buy the most energy-efficient home appliance?'; (ii) 'How often do you do things to reduce energy use?'; and (iii) 'How confident are you that you could use less energy than now?' To construct this sub-index, we weight answers on the predetermined response scale by the corresponding percentage for each response in each country. For example, question (i) above was answered on an 11-point scale ranging from 0 ('Not at all likely') to 10 ('Extremely likely'), and we weight these initial values with the corresponding percentages in each country. 11 We follow a similar approach for each of the three questions, and then aggregate the three individual scores to construct the Energy sub-index. To achieve a consistent range of answers across questions and provide scores that are more meaningful, we rescale the values by applying min/max normalization. In all cases, the sub-indices range between 0 (less environmentally friendly responses) and 100 (more environmentally friendly responses).
The Climate sub-index is based on the answers to the following questions: (i) 'Is climate change caused by natural processes, human activity, or both?'; (ii) 'To what extent do you feel personal responsibility to reduce climate change?'; and (iii) 'How worried are you about climate change?' Using the answers to these questions, the Climate subindex is calculated in the same way as the Energy sub-index. Τhe Policy sub-index is calculated using the answers to three questions revealing the extent of public support for policies to reduce climate change: (i) 'Do you favour increasing taxes on fossil fuels, such as oil, gas and coal?'; (ii) 'Do you favour using public money to subsidize renewable energy such as wind and solar power?'; and (iii) 'Do you favour banning the sale of the least energy-efficient household appliances?' For all three questions, responses were given on a 5point scale: 1 = 'strongly in favour', 2 = 'somewhat in favour', 3 = neither in favour nor against, 4 = 'somewhat against', 5 = 'strongly against'. We quantify and convert these responses in the same way as for the other two sub-indices, thereby creating a Policy sub-index ranging between 0 and 100, with a higher score indicating stronger public support for adopting policies to reduce climate change.
Finally, we calculate the overall index, Environmental perceptions, as the average of the Energy, Climate and Policy sub-indices. 12 Figure 3 shows the national mean values of our Environmental perceptions measure. We observe large differences across countries in the public Environmental perceptions of the public, with values ranging from a low of 9.15 in Russia to a high of 81.0 in Germany.

Estimation method
To examine how public perceptions on environmental issues impact on reputational risk, we derive a model that represents reputational exposure rating as a function of perception, firmand country-specific characteristics. In its general form, the model is as follows: Reputational exposure rating = f(Perception, F irm speci||c , Country S peci||c variables) where Reputational exposure rating refers to the reputational risk rating of an individual firm (as discussed above); Perception represents the index ECP or one of the three sub-indices; and Firm specific and Country specific represent the corresponding control variables. Time and industry dummies are also included in our model. We use an ordered probit model due to the ordinal nature of our categorical dependent variable. Our approach is consistent with that used in the credit ratings literature (see e.g. Khatami, Marchica and Mura, 2016;Papadimitri et al., 2020).

Empirical results
Ordered probit results for H1 Table 5 shows the results of the baseline model. In column 1, we investigate the impact of Environmental perceptions while controlling for basic firm-, country-and industry-level factors. In the next columns we present the results for the sub-indices of Energy (column 2), Climate (column 3) and Policy (column 4).
Environmental perceptions appear to have a significantly positive impact on Reputational exposure rating (at the 1% level), and this finding holds when we decompose the overall index into the three sub-indices. Overall, our results show that in countries where the public demonstrates more environmentally friendly attitudes and beliefs, com-panies are more likely to have better reputational risk ratings, and hence lower reputational exposure, providing support for H1a. One potential explanation is that, consistent with systems theory, these firms' managers carefully analyse the operating environment, including public perceptions, and adapt their practices accordingly (Logsdon and Yuthas, 1997). Therefore, this pressure from the public incentivizes firms to advance their frontier in relation to ESG issues, eventually meeting the expectations of stakeholders and lowering their reputational exposure.

Ordered probit results for H2
Up to this point, we have used energy, climate and the adoption of relevant policies to highlight public perceptions of environment-related issues. Such beliefs may play the role of informal institutions, since firms may be subject to different levels of institutional pressures. However, as discussed, recent research suggests that a company's ability to interpret the public perceptions that drive social institutional pressures, and react accordingly, depends on certain board characteristics.
For instance, foreign chief executive officers (CEOs) may perceive and interpret public pressures differently from local CEOs if they come from a society that does not share the same public aspirations. Consequently, external institutional pressures and CEO nationality may interact to affect corporate reputational exposure. To account for this, we introduce the dummy variable Local CEO that takes the value of one for a local CEO and zero for a foreign CEO, and we interact it with the indicators of public perceptions. We present the results in Table 6. In general, the interaction term is insignificant, failing to provide support for H2.
In addition, based on previous studies on top management teams, we examine the influence of different kinds of board demographics such as age, qualifications, nationality mix and gender. The reasoning is in line with upper echelons theory, which states that organizational outcomes related to strategic choices and performance are partially predicted by the background characteristics of top executives (see e.g. Bromiley and Rau, 2015;Carpenter, Geletkanycz and Sanders, 2004;Hambrick and Mason, 1984). More specifically, Hambrick and Mason (1984) assert that observable characteristics of top managers, such as their age, The results in Table 7 partially support upper echelons theory, with a significantly positive coefficient on the interaction term of Board age diversity with Environmental perceptions, Climate and Policy. Consistent with H2, these findings imply that external institutional pressures and the age diversity of board members may interact to affect corporate reputational exposure. This finding might be explained by the wider range of ideas and perceptions among directors who differ in age, which proves helpful when a firm struggles to find the seminal work of Hambrick and Mason (1984), such as race and gender diversity (Richard et al., 2004). This is why we also consider such attributes in our analysis. solutions for compliance issues (Kumar, 2020), subsequently resulting in lower reputational exposure. This should not be surprising as studies have suggested since at least the late 1970s (e.g. Buttel, 1979) that age is closely and consistently related to attitudinal indicators of environmental concerns. 15 Most importantly, the literature suggests closures (Fernandes, Bornia and Nakamura, 2019) and environmental compliance initiatives (Kumar, 2020). Finally, we find no evidence of a conditional effect from the remaining board demographics (results untabulated), and thus we fail to find further and strong support for upper echelons theory.

Additional analyses
In this section, we present further analysis of: (i) alternative CSR measures; (ii) additional countrylevel attributes; (iii) altered samples; and (iv) endogeneity. We discuss these tests in turn.

Alternative CSR measures.
Various studies point to a positive relationship between social performance and better reputation Pavelin, 2004, 2006). As discussed in Soleimani, Schneper and Newburry (2014), since reputation refers to public perceptions of the firm, CSR reporting and participation in voluntary initiatives are particularly relevant because they provide highly visible signals of commitment to CSR. Additionally, RepRisk mentions that its ratings take into account various allegations related to social issues. We therefore introduce two more CSRrelated variables. The first is a dummy variable, CSR sustainability index, that takes the value of one if the company belongs to a specific sustainability index, and zero otherwise. The second is a dummy variable, Sustainability reporting, that takes the value of one if the company publishes a separate sustainability report or a section in its annual report on sustainability, and zero otherwise.
The results in Table 8 show that both CSR dummies enter our regressions with an insignificant coefficient; the main results do not change.
Additional country-related controls. We further account for several country-related factors that can influence reputational exposure. Using firmlevel data from the WEF database, Walsh et al. (2009) find that customer satisfaction and trust significantly impact on corporate reputation. In the absence of firm-level customer data, we control for buyer sophistication at the country level. This indicator reveals the extent to which buyers base their purchases on sophisticated performance attributes and not only on the lowest price. The results in Table 9 confirm the findings of previous studies in that Buyer sophistication significantly affects Reputational exposure ratings. The effect is positive and statistically significant at the 1% level (except in the model with Policy -column 4 -where it falls to 5%). Hörner (2002) built a theoretical model that shows how competition generates reputationbuilding behaviour. For example, having a competitor in the market may allow consumers to cred-ibly punish one firm's dishonest behaviour, thus raising the importance of building and maintaining reputation. We control for the impact of local competition on reputational exposure using an index of the intensity of competition in each country. We find that Local competition is positively associated with Reputational exposure ratings at the 1% level (columns 5−8), and the environmental perception variables continue to have statistically significant coefficients at the 1% level. Additionally, we control for the extent of cooperation in labouremployer relations. Helm (2011) outlines the important role of employees in reputation building: 'employees can directly or indirectly, voluntarily or involuntarily, affect reputation by any act that is transmitted to, and communicated by, external audiences who evaluate corporate conduct' (p. 658). Miles and Mangold (2014) also highlight that in the era of the Internet and social media, the voice of employees can either enhance the organization's public image or be a ticking bomb with adverse effects on corporate reputation. We find that Labour-employer cooperation is associated with lower reputational exposure (statistically significant at the 1% level: columns 9−12). Additionally, we control for the impact of poverty using the percentage of the population living below national poverty lines (WEF). The World Economic Forum (2012) outlines that in most countries, improvements in economic living standards are being accompanied by increases in political and civil rights such as freedom of speech, assembly and beliefs. This might have further implications for the priorities, attitude and reactions of citizens in wealthy and less wealthy countries. For example, Lo (2016) finds that the citizens of wealthier societies are more strongly motivated to take environmental action than the citizens of lower-income countries, although at the same time they are relatively less likely to perceive the harmful impacts on the environment as very dangerous. Therefore, the expectations of the citizens and the corporate reactions may differ by level of income, with implications for organizational fit and subsequently corporate reputational exposure. The results in Table 10 reveal a negative relationship between Poverty and Reputational exposure ratings, implying that in countries with a high percentage of poverty, companies' reputational exposure is also higher. The main results hold.
Given that the Reputational exposure ratings indicator captures negative incidents, criticism and controversies reported in the media, it is plausible that the level of dissemination of such incidents in a country may drive our results. While it is not possible to examine the exact content reported, we attempt to account for the dissemination of information in a country with a proxy that captures the breadth of print and broadcast perspectives (Press). We explore the extent to which this measure moderates the impact of public perceptions on companies' reputational exposure. The coefficient of the interaction term is insignificant or only marginally significant, revealing a rather weak role in further explaining the aforementioned relationship. Nonetheless, when Press enters the regression as an additional control variable, the coefficients of the key variables of interest remain intact in both sign and significance. We do not tabulate these results to conserve space, but they are available upon request.
We also test how differences in reputation across countries could be explained by variation in the level of institutional development (Deephouse, (3) (8)

Notes:
Ordered probit results from Eq. (1) when enhanced with additional country-level controls. Columns 1, 5 and 9 report results when the key independent variable is the overall ECP measure; columns 2, 6 and 10 when the independent variable is the Environmental sub-index; columns 3, 7 and 11 when the independent variable is the Climate sub-index; and columns 4, 8 and 12 when the independent variable is the Policy sub-index. All financial variables are winsorized at the 10th and 90th percentiles. The variables are defined in Table 1. Robust standard errors are reported in parentheses. * * * p < 0.01. * * p < 0.05. * p < 0.1. Notes: Ordered probit results from Eq. (1) when enhanced with additional country-level controls. Column 1 reports results when the key independent variable is the overall ECP measure. Columns 2-4 report results when the overall ECP measure is decomposed into the Environmental, Climate and Policy sub-indices, respectively. All financial variables are winsorized at the 10th and 90th percentiles. The variables are defined in Table 1. Robust standard errors are reported in parentheses. * * * p < 0.01. * * p < 0.05.
Newburry and Soleimani, 2016). As in past studies, we use the World Bank's WGI as a measure of each country's overall institutional attainment. The interaction terms of WGI are positive and statistically significant at the 5% level (except for Policy), indicating that the impact of environmental perceptions on the reputational risk rating is enhanced in countries with high (compared to low) levels of institutional development. The results are reported in Table 11.
Altering the sample. As discussed above, our main dependent variable has no time-series varia-tion. Our panel-setting approach is consistent with many studies examining the impact of national culture on firm outcomes, under the assumption that such social norms change little (if at all) over short time periods (Chen et al., 2015;Deephouse, Newburry and Soleimani, 2016;El Ghoul and Zheng, 2016). Still, running the regressions with firm-year observations over 4 years could artificially increase the power of the tests. Therefore, we re-estimate the baseline regression using values averaged over the 5-year period, giving one observation per firm. Additionally, we re-estimate our 24 C. Gaganis et al.  Tables 12 and 13 show that our findings are robust to these alternative estimations.
Furthermore, to ensure that the results are not driven by a single country, we estimate our regression by excluding the UK, as the dominant country in our sample (34% of observations). As reported in Table 14, we find no difference in our baseline model. Given the nature of the environmental perception variables, we test a revised specification with standard errors clustered at the country level. As Table 15 reports, the main results hold, with the environmental perception variables being statistically significant for all model variants at the 1% level.
Endogeneity. A potential endogeneity issue clouds the interpretation of our main results. Although reverse causality is unlikely to be a major issue in our setting, public beliefs and views are arguably influenced by the behaviour or actions of local firms (e.g. the BP oil spill of 2010). In addition, endogeneity could potentially be linked to a spurious relationship due to omitted variable bias and/or measurement error.
To be more precise, a spurious relationship occurs when a third variable creates the appearance of a relationship between two other variables, but this relationship disappears when that third variable is included in the analysis. For example, one may argue that country-level national culture can simultaneously drive both public perception (independent variable) and firm reputational exposure (dependent variable), thus creating a spurious relationship. To account for this, we include national culture as an additional control variable in the baseline specification. We consider all six dimensions identified by Hofstede (1980), Hofstede and Bond (1988) and Hofstede, Hofstede and Minkov (2010), namely power distance, individualism (versus collectivism), masculinity (versus femininity), uncertainty avoidance, long-term orientation (versus short-term orientation) and indulgence (versus restraint). Since we are interested in controlling for these aspects of national culture rather than testing some hypothesis about their potential impact on corporate reputation exposure, and given the potential correlations between the six dimensions, we resort to principal component analysis. This results in two components with eigenvalues higher than one, explaining in total 0.7684 of the variance. The inclusion of these two cultural components in the regression does not alter our main findings. Thus, it appears that our results are not driven by spurious correlation due to the omission of national culture. The results are reported in Table 16.
As a further test, to isolate the exogenous component of our endogenous variable and alleviate any remaining concerns about public beliefs being influenced by firm behaviour, we re-estimate (1) with an ordered probit regression that allows for endogenous covariates. 16 We propose that the disaster intensity measure for temperature extremes from Felbermayr and Gröschl (2014) can serve as an appropriate instrument. Felbermayr and Gröschl (2014) (Hamilton and Keim, 2009;Howe et al., 2013;Leiserowitz, 2006;Weber, 2016). Thus, this instrument should satisfy the relevance requirement of an instrument. Columns 1−4 in Table 17 report our findings, with Panel A presenting the second-stage results and Panel B the first-stage results. The key variable of interest (Environmental perceptions) retains its sign and significance, and the results remain intact for the three sub-indices of Energy, Climate and Policy. We also find that the instrument Temp is positively related to all four environmental perception variables at the 1% level, thus confirming its impact on the first-stage regression-dependent variable. Given the lack of a formal set of tests to confirm the validity of the instrument, we re-estimate the first-stage model and examine the likelihood ratio chi-square statistic. We obtain a value lower than 0.001, which confirms the significance of the model. Finally, in untabulated results, we also estimate our model using a two-stage least-squares (2SLS) linear model. Although the 2SLS method disregards the ordinal nature of

Discussion and conclusions
Corporate reputation has received substantial attention in the management literature. However, little is known about the factors that drive reputational exposure across countries. Based on insights from institutional theory, which has been widely adopted to explain how firm behaviour is being driven by institutional pressures, we aim to fill this gap. We focus on public perceptions of environment-related issues like energy, climate and the adoption of relevant policies. Such beliefs may play the role of informal institutions, since societal expectations regarding organizational behaviour are possibly the most influential environmental forces. For example, as discussed in the latest Authenticity Gap Report by FleishmanHillard Fishburn (2019), campaigns such as Extinction Rebellion, widespread climate protests and grow-ing public concern about single-use plastics and irresponsible energy have driven climate change higher on the global agenda. At the same time, attitudes are driven by a greater emphasis on societal purpose and internal scrutiny from the media, and while consumers do not expect companies to fix everything, they will scrutinize them to make a positive difference on the issues under their control. This means that companies nowadays operate within a fast-changing, possibly extreme, informal institutional environment that poses challenges for the management of value and reputation, and subsequently the corporate governance of firms. Using a sample of 643 firms from 19 European countries over the period 2015−2018, we find that more environmentally friendly public perceptions result in lower reputational exposure. Drilling down into public perceptions, we find that this result holds when we disaggregate the overall index of environmental perceptions to sub-indices of public opinions on energy, climate and the introduction of related policies. The results are robust to various firm-level and country-level control variables, and to techniques that address potential endogeneity bias. One potential explanation of our findings is that environmentally friendly societal beliefs put pressure on managers to adopt proactive environmental strategies and achieve a fit between corporate policies and public expectations. In turn, this results in lower reputational exposure. We also find that age diversity in the board of directors moderates the association between environmentally friendly public perceptions and reputational exposure.
These findings have implications for both firm internal stakeholders, and policy-makers and other external stakeholders. First, our study documents for the first time in the literature that public beliefs regarding environmental issues have an important impact on firm reputational exposure. Policy-makers could take this into account in designing policies for environmental and social issues. Additionally, for governments and regulatory agents, this means that one way to strengthen the effect of public pressure on firm behaviour is to raise managerial and shareholders' awareness of this relationship. Environmental regulations could have a complementary role to public pressure, and they could also be complemented with education programmes and informative campaigns for managers and shareholders. For example, the Fleish-manHillard Fishburn (2019) report reveals that 79% of surveyed consumers are concerned about environmental issues and 59% expect companies to take a stand on climate change, ranking the protection of the environment in consumers' top three expectations. At the same time, 84% of the studied firms experience a gap between people's expectations of them in caring for the environment and people's actual experience of what they are doing. Therefore, the report highlights that organizations should take steps to manage together their brand and reputation, and that failure to communicate effectively on the topics that are of interest to the public and consumers means that they risk alienating their customer base. For example, 80% of consumers are prepared to stop using the products and services of a brand if its response to an issue does not support their personal views. We believe that one potential reason explaining this gap is that while many corporate leaders understand the key role of business in tackling climate change, they also believe that pursuing a sustainability agenda runs counter to the wishes of their shareholders (Eccles and Klimenko, 2019). However, after interviewing 70 senior executives at 43 global institutional investing firms, Eccles and Klimenko (2019) argue that this perception is outdated, with ESG running behind in the agenda of these executives. This is confirmed in a survey by Bank of America Merrill Lynch, which reveals that US executives underestimated substantially the percentage of their company's shares held by firms employing sustainable investing strategies, with the average estimate being 5% compared to an actual figure closer to 25% (Eccles and Klimenko, 2019). Appropriately designed programmes and campaigns could increase awareness about these issues, align the interests of all parties and subsequently decrease reputational exposures.
Second, given the relationship between reputation and stock prices established in past studies (Gregory, 1998;Raithel and Schwaiger, 2015), prospective international investors could also consider cross-country differences in public environmental perceptions as these might impact corporate reputation exposure, with subsequent implications for the value of their investment. For example, a recent report by Bank of America Merrill Lynch (2019) highlights that ESG-related scandals can mean huge losses for companies and investors, mentioning that 'Major ESG-related controversies during the past six years were accompanied by peak-to-trough market capitalization losses of $534 billion for large US companies. Loss avoidance is key for portfolio returns over time' (p. 1).
Third, our research brings together the literature on informal institutions and corporate governance. While corporate governance researchers have been debating the impact of board characteristics on firm performance and reputation, our results show that most of the board and corporate governance characteristics do not have a moderating role in the relationship between the informal institutional pressures of public environmental perceptions and corporate reputational exposure. In more detail, board qualifications, board nationality mix and board gender mix do not have a moderating role, with age diversity being the only board characteristic that matters. Existing shareholders could take these findings into account when adopting corporate governance mechanisms. Fourth, our results have shown that the impact of environmental perceptions on reputational risk rating is enhanced in countries with high (compared to low) levels of institutional development. This can have implications for all the above stakeholders. It will also have implications for firm managers, as it shows that the impact of public perceptions on reputational risk is not uniform across countries, and managers will have to take such country-specific characteristics into account while managing reputational risk.

Limitations
Our study is not without its limitations. First, data availability restricted the analysis to a 4-year period. Future studies could explore if the results hold over longer time periods. Second, public perceptions on environmental issues may vary within a country. While data are currently available only at the country level, future research could possibly consider regional public perceptions and their impact on the reputational risk of locally headquartered corporations. Finally, future research could possibly survey executives to reveal their views and strategic decisions in response to the public's environmental perceptions.