Corporate risk reporting about Brexit as political communication

How can we account for differences in the extent of risk disclosure among companies? The paper expands the existing explanations by claiming that corporate risk reporting is not just financial but also political communication. The presented empirical analysis of how corporations disclose Brexit-related uncertainties suggests that risk reporting is a part of a company’s holistic conversation with multiple audiences in society (e.g. politicians, regulators, journalists and customers) and might have well-targeted but also unforeseen effects on each of them. The quantity and quality of risk disclosure can be explained – among other factors – by the extent to which companies want to participate in public discourse and wish their opinions on a particular political issue, such as Brexit, to be heard. In other words, risk reporting is a part of ‘the politics of expectations’ which should be investigated in its own right.


Introduction
In the last decade, economic sociologists have been paying an increasing attention to radical uncertainty and expectations as 'imagined futures'. This debate highlighted the necessity to better understand expectation formation and the diverse roles that expectations play in public spheres of various kinds. Indeed, there are 'the politics of expectations' (Beckert & Bronk, 2018, p. 28) which should be investigated in its own right: How do views of the future influence economic and political action? Which images of the future dominate the discourse? Which narratives stick with economic actors and which disappear unnoticed? How do people become convinced? How do expectations project economic incentives and result from the careful sorting of images in power games? These questions are now recognised to be central for politicians, company managers, central bankers and other practitioners and regulators (Beckert & Bronk, 2018).
In the paper, I use corporate risk disclosure (CRD) as an example of a public sphere and analyse how companies are involved in public claim-making about controversial economic issues. While delivering guidance concerning expected growth, new products, forthcoming acquisitions, etc., corporations participate in public economic debates: they disclose their expectations and comment on political and economic events that influence their business. In particular, I focus on the communication of the risks and uncertainties associated with Brexit.
Brexit is an uncertainty that largely evades companies' control and represents a challenge as to how to meaningfully report about it. Curiously, uncertainty has not been an issue in its own right in the CRD literature: all uncertain future issues have generally been treated as 'risks'. Still, Brexit is a case in which corporations are required to communicate uncertainties (Bloom et al., 2018), and they do it in different ways and styles. Analysing how companies in two UK FTSE 100 sectors (banking and pharma) communicate the insecure Brexit prospects, the paper aims to find an explanation for the differences in the amount and quality of their risk reporting. While existing studies primarily tried to relate the quantity of provided risk disclosure to a specific company's characteristics such as size, industry and governance structure, for example, a company's ownership and the composition of the board of directors (Elshandidy et al., 2018), I suggest considering risk reporting not as a narrow, purely financial issue but as a part of the broader public debate in society. While firms simultaneously talk to various audiences -shareholders, journalists, NGOs, governments, etc. -in many voices, the awareness of being part of the societal discourse motivates them to talk or keep silent about particular aspects of their future. By providing the empirical picture of CRD about Brexit, the paper at hand shows how the involvement of firms in national public debates simultaneously restricts or enables their risk communication with markets. In particular, the paper demonstrates that, while UK banks wanted their voices against Brexit to be heard from the onset, the pharma sector was rather quiet. This is reflected in the different amount and quality of information provided on Brexit and the different styles of risk disclosure in both sectors. Corporate communication about expectations and risks is highly political -there are specific 'politics of expectations' that should be better understood. This is a new and under-researched aspect of CRD to which the paper pays particular attention.
Furthermore, the paper addresses recent concerns about the diminishing utility of the information that companies provide to professional and amateur investors. Despite calls for more detailed reporting and related regulatory changes towards more transparency, the relevance of corporate financial communication seems to be lost (Lev & Gu, 2016). The resulting insufficient investor awareness about companies' futures can lead not only to individual misinvestments but also to the misallocation of capital in the economy as a whole. Thus, a better understanding of the determinants and restrictions of corporate financial communication is very important for the 'health' of capital markets. By discussing accounting communication as a part of the holistic communication strategy of a company, the explorative paper at hand expands the existing, primarily quantitative, studies on risk reporting and points towards an additional explanatory factor that might help to understand the differences in the quantity and quality of CRD.
The remainder of the paper proceeds as follows. In Section 2, I show how researchers have conceptualised and analysed risk reporting so far. I particularly point out the general neglect of the uncertainty discussion in the CRD literature and highlight the latest trends in the analysis of financial communication as a part of the broader public debate. Section 3 explains the data and the methodology. In Section 4, I demonstrate empirically how financial communication around Brexit has been organised in two sectors: banking and pharma. Section 5 summarises the findings and suggestions of the paper.

Understanding risk reporting so far
In the 1990s, the idea that companies should not only report results and provide earnings guidance but also disclose risk information to investors took hold in the UK. The Institute of Chartered Accountants in England and Wales (ICAEW, 1997) highlighted the benefits of forward-looking risk disclosure in companies' annual reports: better estimates of companies' earnings and cash flows by investors, lower cost of capital and improvement of risk management within companies.
As soon as CRD became an established practice, accounting researchers started to investigate whether this practice brought the expected benefits for investors. While predominantly applying content analysis, i.e. counting words and sentences that mention risk in annual reports, researchers investigated the quantity of risk information provided by companies. In order to explain why some firms report more about risks (i.e. mention risks more frequently in their reports) than others, the studies on CRD have focused on the role of regulatory requirements (mandatory versus voluntary risk-reporting environment), the governance structure (such as company ownership, the composition of the board of directors, the existence of internal auditing or a risk committee) and other factors, such as company size, industry, etc. (Elshandidy et al., 2018).
This primarily quantitative research on CRD was informed by the transmission model of accounting communication (Brennan & Merkl-Davies, 2018) and the principal-agent theory, according to which companies (as agents) are rather unwilling to disclose information about future risks to investors (as principals) but still do it -often pro forma -in order to comply with regulations and to avoid disinvestment as a punishment for providing wrong or insufficient information. This simple transmission model is monological and one-directional: information is presented by companies in annual reports to just one audience, namely investors, and is passively 'consumed' or ignored by the latter. How much information is provided depends only on the companies, their will and characteristics. The literature on CRD that complies with the transmission model focuses on whether risk information is reported and, if so, which information is reported.
This analytical framework, however, disregards the fact that risk disclosure is primarily about communication and narratives; thus, a meaning-oriented (qualitative) analysis of 'how' and 'why' risk issues are disclosed is also needed (Mazumder & Hossain, 2018). The paper at hand aims to provide this type of analysis.

Corporate risk reporting as a part of a societal discourse
Recent research on accounting communication has started to move from the transmission model towards systematically analysing the ongoing commentary on and negotiation of companies' information by multiple interconnected actors who talk to each other through various media and channels. Brennan and Merkl-Davies (2018) introduced the transactional model, according to which financial corporate communication cannot be reduced to one-way information transmission. Risk reporting is a case where a dialogue with multiple actors shapes their respective practices and ultimately forms audiences for their communications, while providing information and feedback and forming relationships ( Figure 1).
Stories about companies' uncertain futures are created by CEOs, CFOs and investor relations within a company and communicated to the financial public in annual reports but also in press releases, investor conferences, road shows and social media. However, this is not a one-way process: the unique qualitative study by Åhblom (2017) shows in great empirical detail how numerical and non-numerical information provided by companies is shaped by an interactive relationship between shareholders, financial intermediaries and investor relations. Indeed, the company's narrative is constantly challenged and re-interpreted by security analysts and fund managers who give feedback to the management not only by buying or selling the company's shares but also by asking questions and challenging the management in one-on-one meetings, investment conferences and earnings calls. The negotiated and interpreted versions of the company's future become inputs for investors' models, investment decisions and financial analysts' reports.
It is important, however, to emphasise that the transactional concept of corporate financial communication goes beyond the scope of the investment community. Corporations know that their financial communication, including risk reporting, simultaneously addresses many audiences (Åhblom, 2017;Whitehouse & Perrin, 2015). Those audiences are not reached incidentally. By telling a story about the future risks to which they are exposed, companies enter the public sphere where not only shareholders but also competitors, governments, customers, rating agencies, NGOs, journalists and the general public listen to and interpret the message. Seeking to achieve their specific goals, all those audiences read annual reports and search for information in them. For example, representatives of companies' R&D departments read annual reports to be informed about competitors' product pipelines; activists try to obtain an impression about the corporate climate change policy and the CO2 footprint, etc. Thus, while preparing their narratives, companies always keep various audiences in mind and pay attention to the feedback they have received. Indeed, as Bhatia (2012) suggested, corporate annual reports -and I would add other media of communication -comprise at least four different discourses: next to the accounting, financial and regulatory discourses, there is also the 'public relations' discourse which is directed at multiple audiences. Thus, corporate narratives are simultaneously co-shaped by the communicative environment of a company: government statements, mass media reports, economists' opinions and regulatory changes. Importantly, the audiences interact with each other and contribute to 'a struggle over meaning' (Brennan & Merkl-Davies, 2013): security analysts provide feedback to companies; journalists interview analysts and CEOs and transmit corporate messages to the general public; rating agencies derive information from analysts' reports, etc.
Furthermore, the most recent literature encourages researchers in financial communication to engage in the narrative turn and pay more attention to textual elements and rhetoric aspects of various genres such as ad-hoc corporate announcements, annual reports, earnings calls, financial analysts' recommendations, takeover documents, IPO prospectuses and rating decisions (Palmieri et al., 2018).
This wider view of financial communication as an interactive process has not been explicitly recognised in the risk reporting literature so far. However, if we start to consider CRD as a part of the wider public debate that companies lead with society, we could gain additional insights into their motives for disclosing or withholding risk information and for presenting a piece of information in a particular way. There are social processes of controlling the discourse, noticing, ignoring and silent witnessing going on in the communicative networks presented in Figure 1. Companies' involvement in public debate codetermines the style and the scope of their stories. In this paper, I analyse corporate risk reporting about Brexit understood exactly in this sense -as a part of 'the politics of expectations' in the public domain (Beckert & Bronk, 2018, p. 28). Beckert and Bronk (2018) argue that 'the politics of expectations' deserve closer attention under conditions of radical uncertainty. Brexit is an uncertain event which concerns many companies in the UK and thus necessitates communication of the associated risks and possibly catastrophic futures that are unknown at the time when the financial communication is prepared. Directly after the vote to leave the EU, UK companies received the recommendation from the Financial Reporting Council (FRC, 2016, p. 2) 'to consider the consequential risks and uncertainties in the political and economic environment and the impacts of those risks and uncertainties on their business' and to make them a part of their corporate reporting. Accountants and directors were expected to 'fit' Brexit reporting into the existing principal risks framework in annual reports.

Uncertainty and risk in the CRD literature
What is striking about this guidance is that no conceptual distinction was made between risk and uncertainty. This distinction was introduced by Knight (1921) and Keynes (1921Keynes ( /2006Keynes ( , 1937Keynes ( /1973, and has been further developed by some post-Keynesian economists (Davidson, 1988(Davidson, , 1991(Davidson, , 1996Dequech, 2000Dequech, , 2006Dow, 2012;Shackle, 1949Shackle, , 1955Shackle, , 1969. Knight, Keynes and Post-Keynesians distinguished between risk as a situation in which there is some objective basis for determining probabilities of possible outcomes, and uncertainty, in which there is no such basis, even if decision-makers assign subjective probabilities, since the events or actions are not taking place in identical or even similar conditions. Although the importance of uncertainty (as opposed to risks) has been widely discussed in organisation and management studies (Alvarez & Barney, 2005;Townsend et al., 2018), decision making literature (Epstein & Wang, 1994;Gilboa et al., 2008), psychology (Tuckett, 2018;Volz & Gigerenzer, 2012), economic sociology (Beckert & Bronk, 2018) and -most recently -in nonheterodox economics (Akerlof & Shiller, 2009;King & Kay, 2020), this 'uncertainty turn' has not arrived in risk reporting research. Maybe surprisingly, the notion of uncertainty has not been an explicit issue there at all.
Despite calls to invest more work in the discussion of the risk concept used by both companies and investors (Elshandidy et al., 2018, p. 76), not much has been done in this respect. The terms 'risk' and 'uncertainty' are used interchangeably by researchers and regulators. The Institute of Chartered Accountants in England and Wales (ICAEW, 1999, p. 3) defines risk, for example, as 'uncertainty as to the amount of benefits'. Dobler (2008, p. 187, my emphases) adopts a similar term, that of 'uncertainty-based risk', defining 'risk as randomness or uncertainty of future outcomes that can be expressed numerically by a distribution of outcomes'. In a comprehensive review of the literature on the quality of risk reporting, Ryan (2012, p. 296) points out the Knightian distinction between risk and uncertainty and notices that 'most of the empirical literature [...] does not distinguish these notions'.
To my knowledge, the only official document on corporate risk reporting that mentions Knightian uncertainty is the ICAEW study (2011). It highlights that some business risks can be approached by means of probabilistic calculus and scenario analysis as they can be calculated based on an established statistical record (e.g. in insurance). At the same time, the ICAEW study claims that many risks are unmeasurable uncertainties but does not discuss the consequences of this conceptual insight for risk reporting practices.
A more detailed discussion on the distinction between risk and uncertainty would have helped though in the situation of Brexit. 'Brexit provided both a major and persistent uncertainty shock' (Bloom et al., 2018, p. 556) with many unknowns which companies might struggle to quantify and materially communicate. Thus, if we agree that Brexit is not a calculable risk, we might want to analyse Brexit reporting through the lens of uncertainty.
Often, uncertainty is understood in the literature as a limit to information. While focusing on 'whether' and 'what', existing CRD research has touched upon uncertainty exactly in this sense. Dobler (2008) discussed information endowment, i.e. what managers know or can know and whether this knowledge is communicated properly to investors. However, for a deeper discussion of risk reporting about Brexit, the complementary concept of uncertainty might be helpful. Indeed, recent literature in the sociology of economics and finance has highlighted that 'the question of uncertainty is ill-understood if it is solely framed as an issue of knowing or not-knowing' (Tellmann, 2016, p. 64). Uncertainty is not just a given external condition but also a product of various practices such as risk management (de Goede, 2004;Langley, 2013) and risk communication (McGoey, 2009(McGoey, , 2019. By presenting an issue as uncertain, companies can attract or divert attention, define the situation, convey their authority and persuade, i.e. develop a conviction narrative (Tuckett, 2018). In other words, uncertainty might be a useful resource for the communication of companies with their audiences and, importantly, might have performative effects.
Thus, uncertainties such as Brexit might bring about specific styles of reporting. Facing uncertainty, the companies do not just decide 'to disclose' or 'not to disclose' information. They might also refer to it a lot (e.g. mention risks many times in the report) without providing any substantial details about risks' nature, dimensions of risk etc. The CRD literature has noticed symbolic, or boilerplate, disclosure of this kind (Abraham & Shrives, 2014;Linsley & Shrives, 2006), but has not connected it to uncertainty: '[R]isk disclosures which are general or routine may have the appearance of valid disclosures but in actuality no information content is provided and readers do not find them useful' (Abraham & Shrives, 2014, p. 94). Symbolic disclosure is a specific form of silence in annual reports.
At the same time, the literature in management and organisation studies has noticed that -especially under conditions of uncertainty -companies' expectations often take the form of 'symbolic planning' (Clarke, 1999) or 'bullshitting' (Christensen et al., 2019;Spicer, 2013). Often, the reaction to uncertainty is a vacuous, airy and obscure talk. For example, Clarke (1999) claims that when organisations are exposed to social pressure to know the unknowable and control the uncontrollable, they react by means of purely rhetorical 'fantasy documents'. Facing uncertainty, companies do not admit the limitations of their knowledge and become involved in providing an illusion of preparedness and robustness while developing plans that are of rhetorical, or symbolical, but not operational use. Thus, one might expect that -in view of Brexit uncertainty -symbolic disclosure and impression management (Brennan & Merkl-Davies, 2013) dominate. However, this should be established empirically. Thus, I pose the following research questions: RQ1: Is risk reporting about Brexit symbolic or substantial?
RQ2: Are there differences among companies with respect to symbolic and substantial Brexit reporting?
Indeed, there are indications that Brexit reporting in the UK is predominantly of a symbolic nature. The FRC report (2019, p. 3) shows that, in investors' opinion, companies do not provide enough 'detail on the level of preparedness, the current stage of implementation of mitigating activities and numerical breakdowns to help them assess the impact [of Brexit]'. At the same time, a generalised view would not be entirely correct: Both the FRC report and my empirical materials discussed below suggest that some companies have responded by providing greater detail on the implications of various Brexit scenarios for their business, whereas others have stuck with boilerplate pro-forma disclosure. The existing literature does not sufficiently explain the differences in the response to uncertainties such as Brexit in companies' risk reporting. Thus, this paper asks: RQ3: How to explain the differences in companies' disclosure about Brexit?
To introduce a possible approach to an explanation, I would like to point again to understanding uncertainty as a performative resource which might be produced and utilised in companies' communication with various audiences. Insistence on the uncertainty of a situation versus heroic claims about taming it would convey different messages and have different effects on the public. For example, McGoey (2009) describes how executives of a big pharma company, Merck, emphasised that they were uncertain about the results of clinical studies and thus did not feel obliged to disclose them. Communication of the impossibility to manage uncertainty might be as important a strategy as insistence on the ability to 'control the uncontrollable'.
More generally, companies are concerned not only with managing and communicating risks but also with carefully managing risk perceptions. The latter can be a source of uncertainty in their own right. No company can be entirely sure of how its messages will be understood and interpreted by various audiences. Thus, every communication creates performative uncertainty, namely uncertainty about how words influence and shape reality (Austin, 1962). This analytical framework is broader than the impression management concept which highlights just one type of performative effect, such as manipulating and at times deceiving shareholders. The performative understanding of uncertainty, however, suggests that communication effects might be multifaceted and even unknown in advance. There are always 'overflows' (Callon, 1998) which should be investigated empirically, also in the context of risk reporting.
In the following empirical part of the paper, I will uncover those processes by discussing two distinct cases of Brexit reporting in the pharma and banking sectors.

Data and methodology
The empirical materials for this study were collected in two steps alongside the corporate communication chain, as depicted in Figure 1. I focussed on two sectors that are strongly affected by Brexit in their business. Five banks and three pharma companies represented those sectors in the UK FTSE100 index in 2019. The companies were anonymised as BANK 1, 2, 3, 4 and 5 and PHARMA 1, 2 and 3. Clearly, the limited number of firms does not support any generalisation claims but serves the exploratory character of the study at hand well.
In the first step, annual reports for three financial years, namely 2016, 2017 and 2018 (24 in total), were collected and analysed. In each annual report, Brexit references (such as 'Brexit', 'referendum', 'decision to leave the EU', 'the UK's withdrawal from the EU', etc.) were identified using the Nvivo software. The identified statements became the units of analysis for RQ1 and RQ2. They were examined by means of manual content analysis (Smith & Taffler, 2000). I applied the approach suggested by Abraham and Shrives (2014;based on Day & Woodward, 2004) which allowed me to categorise each Brexit-related statement as either symbolic or substantive. Table 1 provides the three-level coding scheme. The statements that are (1) general in nature (not idiosyncratic) and apply to any business or any company within the industry, (2) provide only descriptive (qualitative) information, and (3) remain unchanged over years (disclosure inertia) were classified as symbolic. The company-specific statements that identify a company-specific impact and/or quantify the impact of Brexit and change over time were considered to be substantive. 1 Currently, a number of areas for action have been identified including duplication of release testing and procedures for products based in the EU27 and the UK, transfer of regulatory licences, customs and duties set up for introduction or amendment of existing tariffs or processes and associated IT systems upgrades.
PHARMA 2 Table 2 provides examples of symbolic and substantive statements from the analysed annual reports. This first research step allowed the determination of whether Brexit reporting in both industries was substantial or symbolic, and whether there were differences in the reporting style among the companies.
To address RQ3, in the second step, the scope of materials was broadened in order to capture risk communication in the wide sense ( Figure 1). Semi-structured, in-depth interviews were conducted with investor relations, security analysts, portfolio managers, professionals in corporate communication and a representative of an accounting regulation body. Access to the field proved to be difficult due to uncertainty surrounding Brexit. The fully anonymised list of interviews conducted during the study is provided in Appendix. A semi-structured interview guideline was developed based on the literature presented in Section 2. The open-ended questions in the research instrument were divided into the following main themes: (a) questions about the role and experience of the interviewee; (b) probing questions about the distinction between risk and uncertainty (e.g. importance of quantification, role of narratives); (c) practices and strategies of companies' communication of the unknowns (new technologies, new products, Brexit and any other suggested by the participant); (d) probing questions about the reasons for symbolic and substantial disclosure; (e) questions about various communication channels (AR, face-to-face meetings with analysts, investor conferences, social media, etc.) and audiences.
All interviews were conducted face to face and lasted between 1 and 1.5 h. Eleven interviews were digitally recorded and transcribed in full; one interview was documented by taking extensive notes.
Also, to improve my understanding of the field, I participated -as a researcher -in the IR 'best practice' workshop in London organised by the UK Investor Relations Society.
Finally, to better account for the role of the public in shaping companies' risk reporting, I triangulated the interviews with transcripts of companies' earnings calls and meetings with investors that took place in 2016-2019 (30 in total), written analysts' reports on companies of interest published in 2016-2019 (147 in total) as well as with newspaper articles written by professional journalists.
Transcripts of interviews and written materials collected in Step 2 were coded and categorised using the Nvivo software. In contrast to Step 1, the coding at this stage of research was not driven by theory but focused on identifying and establishing themes generated by the data. This bottom-up coding suited the explorative character of the study and allowed the identification of themes not broadly discussed in the literature such as 'symbolic reporting as a form of political silence' and 'performative uncertainty'.

Substantial versus symbolic risk disclosure in banks and pharma companies
Both the banking and pharma sectors are severely impacted by Brexit. First of all, they are subject to heavy regulatory changes. In the time period covered by this study (2016)(2017)(2018), banks were concerned with the possible loss of the right to serve clients in all EU countries other than the UK, the so-called passports. Also, the after-Brexit status of some derivative contracts was unclear. For the pharma sector, there was strong uncertainty whether medicine tests conducted in the UK cease to be valid in the EU and vice versa, necessitating the duplication of testing and licencing procedures, labs and jobs in the UK and the EU. More generally, the details of the new agreements regarding passporting rights and medicine approvals could not be known in 2016-2018 as they were subject to later negotiations during the transition period. Besides regulation, there were further uncertainties that put pressure on both industries. Banks and their investors were concerned that Brexit will damage the UK economy leading to a recession, causing bad debt charges to increase and interest rates to stay low, and thus, adversely affecting banks' net interest margins.  The pharma industry worried about new tariffs and possible delays at the borders. Table 3 suggests that Brexit reporting in both sectors was overwhelmingly symbolic in 2016-18 (RQ1). This applies to both risk sections and other sections of annual reports. Companies focused on disclosing generic (not companyspecific) issues, seldom tried to quantify the impact of Brexit, and often did not change their statements over time.
At the same time, it is evident that both sectors took different approaches to Brexit communication (RQ2). Banks started to actively discuss Brexit and its consequences immediately after the referendum results were published in June 2016, whereas the pharma sector has been very economical with providing information.
Pharma companies claim that their goal is to maintain the supply of medicines to their patients and clients across the UK and the EU and to be prepared for all eventualities, while developing and implementing internal contingency plans. Steering committees or working groups have been established for this purpose. However, the readers of annual reports do not learn which scenarios exactly have been discussed and which numerical assumptions have been made. The mitigation actions include duplication of testing procedures for products in the UK and the EU27 markets, re-design of packaging and labelling, additional inventory builds and securing the supply chain. Pharma 1 quantified the potential cost of the implementation of their contingency plans in 2017, followed by Pharma 2 in 2018. This was nearly only company-specific information available in their reporting. In their reports, pharma companies slightly changed the wording but not the essence of what they reported about Brexit in 2016-2018 so that the information hardly changed over time. All in all, the pharma sector applied a quite similar 'we-are-on-top-of-things' rhetoric. Still, the sentence such as 'over longer term, we continue to believe that Brexit will not have a material impact on our business' (Pharma 1, annual report 2018) might be surprising in view of the potential implications of regulatory changes with respect to medicine trials and licencing, trade and tariff agreements, possible disruptions of the supply chain and access to funding and talents (PWC, 2018).
The discourse developed in the banking sector has principally differed -in its quantity and intensity -from that in the pharma industry. Banks have deliberated in depth on the Brexit uncertainty and its consequences in their annual reports (Table 3). 2 Already in their 2016 annual reports, some of them discussed company-specific measures and announced that they were going to obtain EU licenses and shift some operations, jobs and assets to one of the EU countries to ensure they could continue doing business without passporting rights. In 2017, they started to announce the costs associated with those mitigating actions.
Still, the numbers in Table 3 suggest that banks have by no means completely avoided symbolic Brexit disclosure in their communication. They have also developed stories in the already described 'we don't know what will happen, but we are on the top of things' style. In annual reports, banks declared that they were preparing for all eventualities, conducting stress tests and portfolio reviews. However, in contrast to pharma companies, this preparedness narrative was accompanied by a considerable amount of substantial disclosure.
Whereas quantitative research on CRD would stop here and simply report on the differences between two sectors, qualitative analysis allows us to take the next step and to address an interesting question (RQ3): How can we account for the difference in the extent of risk disclosure in two sectors that are both strongly exposed to Brexit uncertainty? One of the possible answers emerging from the empirical materials collected in Step 2 of this research is: the quantity and quality of risk disclosure can be explained -among other factors -by the extent to which companies want to participate in the public discourse and wish their opinions on a particular political issue to be heard. There is a social aspect to Brexit (non-)communication: risk reporting is a part of a company's holistic conversation with multiple audiences in society and might have well-targeted but also unforeseen effects on each of them.
Risk communication is not just financial but also political communication. What companies report about their risks and mitigation actions can be heard by politicians, regulators, journalists and customers. If a company announced that Brexit required high-cost adjustment measures, e.g. the relocation of the headquarters to continental Europe and severe job cuts in the UK, it would be making a political statement that Brexit is costly, putting off the customers who voted for Brexit and annoying some politicians. On the contrary, by claiming that its business will be unaffected by Brexit (while having UK exposure), the corporation might co-create the narrative that the UK market and maybe the country as a whole are unimportant (as they are too small to have an effect) and cause damage to its own image (as being unpatriotic, etc.). Although the general rule is 'corporations do not do politics', companies often cannot -or do not want to -avoid taking a stance in political affairs (Korschun & Craig Smith, 2018). To elaborate on this, let us turn to our two examples.

Pharma firms
The pharma firms did not volunteer comments on Brexit or go into any details, not only in their annual reports but also in their communications through other channels, for example in earnings meetings and telephone conferences with financial analysts. An investor relation manager of Pharma 2 confirmed that Brexit would not be on the presentation slides prepared for a road show in 2018: 'We wouldn't put Brexit down on paper or a slide outside of the annual report' (Interview 4). There seems to be a kind of filter between the company's backstage and front stage: you know, Brexit is a . . . where we kind of, you know . . . we did lots of scenario reviews as to what might be a problem on things internally and then we decided in the end that there are only one or two which might be a problem externally. (Interview 4) Pharma 2 mentioned Brexit in a conference call on the 2018 results, repeating primarily the information delivered in the annual report. In total, in 11 conference calls of pharma companies with investors, Brexit was briefly referenced twice by the companies themselves in the management introduction. No questions about Brexit were asked by analysts during the calls. More importantly, out of 54 analyst reports on the three pharma companies under scrutiny, only two documents mentioned Brexit, presenting it as a factor that has a positive impact on currency and supports earnings.
Thus, the pharma companies have not developed an extensive discourse about the Brexit implications with the investment community. They comply with symbolic disclosure which allows them not to reveal the potential shortand long-term impact of Brexit on their business. This symbolic disclosure is characterised by a specific treatment of uncertainty, namely uncertainty as limited knowledge. For example, referring to the 'unknown unknown' element of Brexit, Pharma 2 states in the annual report: 'Until the Brexit negotiation and parliamentary ratification processes are completed, it is difficult to anticipate the potential impact on [Pharma 2's] market share, sales, profitability and results of operations.' Also, the investor relation manager of the company said in an interview: we export quite a bit from the UK. Now we don't know if that's going to be subject to tariffs, we don't know if it's going to be subject to customs inspections at the border. So, there's a great deal of uncertainty, because we don't know what the government has decided or negotiated with the EU. (Interview 4) This particular treatment of uncertainty as an epistemic limitation allows them not to disclose much information of a substantial nature. The gist is: 'Too many parameters about Brexit are unknown, so there is not much to say.' This stance also allows them to minimise unwanted or uncertain effects of communication.
'Companies avoid talking about uncertainty in order not to put concerns in the minds of investors', as a representative of the accounting regulatory body suggested (Interview 5). It is why not only Brexit uncertainties, but -as the CRD literature notes -also many other risks are generally under-communicated: there is performative uncertainty, that is, uncertainty about how what is said and presented is perceived by various audiences.
Due to the strong hybridisation of audiences (as highlighted in the IR best practice workshop in London), companies must consider the effects of their communication about risks and uncertainties not only on professional shareholders but also on the general public, including journalists and lay investors. The obligation to be as transparent as possible about all eventualities coincides with the performative aspect of communication, that is, the danger of making things real by sheer talking. Corporations might be very economical with the publication of their scenarios and projections because they internally calculate -among other scenarios -the worst case, and factor in severe disruptions, and some audiences might 'read' the worst-case scenario as a possibility, see it as what a company thinks will happen. Thus, the moment you disclose something to a . . . an audience like analysts, who will get it, at the same time you will have a class of generalists who will misinterpret it. (Interview 11) The head of a business communication company provided two examples. First, he described how 'during the Olympics, we had plans for kidnappings. We don't want to publicise this. [...] We don't want even to entertain the notion of the possibility of a kidnapping.' Second, he made a comparison to NATO: NATO calculates various scenarios but discloses only the scenarios about which it wants to inform the public. That is, these scenarios are communicated and, upon entering the public sphere, automatically become considered as being likely; the others are silenced (Interview 11). Keeping these communicative impediments in mind, symbolic disclosure indeed represents a way of avoiding performative uncertainty, and this is exactly what pharma companies achieve with their risk reporting.
These efforts not to say too much are reinforced by the other important participant in the communication chain, namely the government. As 'The Economist' (2018, p. 23) reported, drug companies working with the government have recently been prevented from talking about the planning needed, by strict gagging clauses. That may make it easier for commercially sensitive information to be shared between government and industry, but it limits the scope for scrutiny or reassurance.
The government wanted to avoid performative uncertainty with respect to Brexit, i.e. to avoid even the possibility of misperceptions and panic precluding pharma companies from sharing scenarios and communicating, for example, possible disruptions of the supply of vital medicines. Of course, the signed non-disclosure agreements are not directly related to risk disclosure; still, they might impede companies from revealing some scenarios. Here, we see how corporate narratives might be co-shaped by non-financial agents who steer them towards non-disclosure and influence the companies' willingness and ability to participate in the public discourse.
More generally, symbolic disclosure allows corporations not to take a political stance. By disclosing the potentially disastrous Brexit scenarios and the related contingency plans, companies would convey a position in the public debate, that is, pro-Brexit or anti-Brexit, or at least their statements could be interpreted as such. Pharma companies clearly try to avoid this. Although their managers have been increasingly pushed, in their public appearances, to provide more clarity on the future UK-EU relationship regarding medical trials and approvals, trade terms, safety around products and on the role of the European Medicine Agency (FT, 2018a(FT, , 2019a, they have preferred not to make corporate financial reporting a tool with which to express their concerns.

Banks
Banks have developed a more pro-active public discourse about Brexit, not only in annual reports. Bank managers frequently addressed the Brexit issue in earnings calls and investor meetings. At the same time, security analysts also actively asked questions about Brexit on nearly every occasion and insisted on receiving clear answers ('And then there's Brexit I mean that comes up in every single meeting', as one analyst noted during the earnings conference of Bank 3). Out of 93 analyst reports, 47 discuss the consequences of the referendum; 10 of them are special Brexit updates. Importantly, next to the 'we-are-on-top-of-things' communication, banks have developed a different story that highlights how Brexit harms them individually and the banking industry in general. They have used their financial reporting to convey to a broader public, including government, that Brexit will cause irreversible damage to the City, that a no-deal will be a disaster and that the responsible UK negotiators should consider banks' interests.
By doing so, banks have referred to Brexit uncertainty not just as an epistemic issue. While highlighting that economic, political and regulatory environments are uncertain because of Brexit and things are partly out of control, banks have utilised uncertainty as a performative resource. Instead of constantly dressing up their outlook (as the impression management concept would presuppose), they have rather dressed down their prospects and have readily provided bad news while clearly referring to Brexit uncertainty. They have described how Brexit jeopardises their growth, costs and NIM (that is, the net interest margin which is the core measure of a bank's success).
For example, Bank 5's manager claimed in the earnings call that they cannot grow business because of Brexit uncertainty: If you are a retail and commercial bank like we are, the preparation for an event like the possibility of a hard Brexit cannot be then 3,6,9 months in advance because we are an accrual bank. So, the stock you have versus the new business is incredibly higher. So, I have been repeatedly asked over the years, why don't you grow more [...]? And that is the best preparation for an unforeseen event such as a cliff event.
Brexit was also discussed as a reason for the low NIM of Bank 1. When analysts asked whether there was a hope for NIM improvement, the manager responded: 'And I got to tell you, for the sake of fixing NIM for you guys, I am not going to change it until we've gone through Brexit' (Bank 1). Importantly, the bank managers' comments do not stay within the narrow investment community but become part of a broader Brexit discussion, for example, in the media. An important audience is journalists who transmit corporate narratives to a broader public (Figure 1). For instance, the Evening Standard (2019) immediately publicised the statements of Bank 2's CEO on the risks for the UK economy arising from Brexit, particularly caused by delays in corporate clients' investment decisions. In the earnings call, the CEO argued: But I don't think I'm alone in saying that the political uncertainty around Brexit has gone on far too long. Our corporate clients are pausing before making financial decisions. And this is, of course, damaging the U.K. economy and will affect our income performance, clearly conveying his discontent to the public. On this occasion, the newspaper also interviewed security analysts who cover UK banks and attended the earnings conference, allowing their voice to also join in the public debate.
A very telling story also refers to the exceptional charges banks announced in 2018 in order to cover a rise in credit defaults. According to accounting rules, banks must calculate potential credit losses based on several economic scenarios they develop internally. Importantly, the adjustment of credit loss charges is subject to management discretion: 'a management overlay to economic uncertainty' (Bank 3). Bank 3 (my emphases) reported in its annual report that [a]t 31 December 2018, management determined that its view of the distribution of possible economic outcomes in the UK was better reflected by using three additional downside scenarios in place of the UK consensus downside scenarios. This resulted in the recognition of additional impairment allowances of $410 m compared with those implied by consensus forecasts, an increase of $165 m in the adjustment to the consensus position compared with 1 January 2018, to reflect the increased level of economic uncertainty in the UK.
Certainly, accounting rules and the bad experience during the global financial crisis a decade ago forced banks to be cautious. At the same time, dressing down the outlook so strongly also suggests a message of a non-financial kind. Indeed, while explaining the exceptional charges for the future credit defaults, the manager of Bank 1 made a clearly political statement in the analyst call after the publication of the 3Q2018 results: During the quarter, I think that you probably agree with me that the risk of Brexit has increased. We are not at a stage we want to be at as a nation.
Importantly, the story about exceptional charges was heard not only by investors but also by the general public. For example, journalists picked up the story: the FT published an article on banks' provisions at the end of the reporting season, highlighting the total provision sum of three banks to be £500 m (FT, 2019b) -an impressive number -and transmitting the worries of bank managers to the readers. Also, the costs of preparation for Brexit were discussed in the FT, for example, in an article 'Big banks boost Brexit budgets: Spending spirals as uncertainty persists over how Britain's EU departure will pan out' (FT, 2018b), implying that surviving under Brexit would lead to hugeand rising -spending in the banking industry. Brexit and its costs to the economy have been a notorious topic in the media, fed partly with banks' reporting (e.g. The Economist, 2018).
Thus, uncertainty communication and risk disclosure can become a tool to convey a clear message in a public debate. Take another example: Bank 1's representatives expressed themselves openly in the a sell-side breakfast meeting, commenting on their 2Q 2018 figures (my emphases): No-deal Brexit -in terms of advice for the industry, our view unfortunately all along was we should have to plan for a no-deal Brexit. It's the only planning assumption we could make and therefore when that will be up and ready, assuming a no-deal, I think we've been very public at urging anyone who will listen to us that a no-deal is not a great outcome for the sector, let alone [Bank 1], and it's a levelling playing field . . . But it won't be good for the sector and industry and it won't be good for European banks, it won't be good for US banks, and it won't be good for Swiss banks. But it's no good for anybody. There's no winner in all of this, so I'd urge policymakers to just be sensible around these things. I think, certainly on the UK side, we're getting more of an audience. They've got to be a little bit careful.
This quote shows that -through this seemingly purely financial communication channel -banks address policymakers and other audiences. Thus, the example of UK banks confirms that risk communication is not an isolated financial affair; the disclosure or silencing of information can be motivated by a company's wish to be heard in the broader public debate.

Conclusion and ideas for future research
The presented research widens the scope of the discussion on corporate risk reporting. It emphasises that there is no unique style which companies apply to risk disclosure: risk reporting can be more or less extensive, symbolic or substantial. While acknowledging the findings presented in the accounting literature that these differences in quantity and quality of CRD are determined by companies' characteristics such as size or governance structure as well as by the regulatory environment, the paper introduces an additional determinant: CRD is a part of the societal 'politics of expectations' which co-explains what is said and silenced in risk reporting. By disclosing their views about the future, companies participate in public debates, a fact that deserves a further, more detailed examination.
Against this backdrop, I would suggest that more qualitative analysis, especially on risk perception and performative uncertainty, is needed: With which voices do companies speak to various audiences and how do those voices interrelate? At the same time, the influence of the different links in the communication chain on how firms disclose uncertainties should be investigated in its own right. It might be particularly interesting to connect CRD and media coverage -an analysis already initiated in this paper.
From a more conceptual perspective, the relevance of the risk-uncertainty distinction for CRD should be investigated: Do companies make such a distinction in their reporting practices? Are there 'known' risks at all? Does the transmission model hold for risks? The research in this paper suggests thatdue to the importance of various audiences in the co-production of risks and uncertainties -the transmission model should be criticised more radically, and the transactional model further developed.
All of these future research steps might enhance the 'relevance lost' debate on CRD in accounting. They would allow us to leave the dominant 'corporations -accounting users' research axis and open the discussion to wider audiences and to include social aspects such as the mutual interrelations of corporations and their audiences, feedback loops, silence and power in financial communication. These insights might help standard-setters and regulatory bodies to better understand the limits of risk reporting: The calls for greater transparency may prove futile not due to companies' deceptive intents but because there are other incentives to be eloquent or keep silent, namely incentives that relate to the structure of the communicative chain as a whole.