The Influence of News Coverage on Humanitarian Aid: The Bureaucrats’ Perspective

ABSTRACT We examine if and how news coverage influences governments’ humanitarian aid allocations, from the perspective of the senior bureaucrats involved in such decision-making. Using rare in-depth interviews with 30 directors and senior policymakers in 16 of the world’s largest donor countries, we found that the majority of these bureaucrats believed that sudden-onset, national news coverage can increase levels of emergency humanitarian aid allocated to a crisis. They said that this influence operated by triggering other accountability institutions (the public, civil society, elected officials) who put pressure on aid bureaucracies to announce additional funding. However, these practitioners claim that annual humanitarian aid allocations—which are much larger—are unaffected by news pressure. Intriguingly, we also find that many respondents interpret a lack of news coverage as grounds for increasing their annual aid allocations to what they call “forgotten crises”. We argue that “bureaucratic mediatisation”, rather than the “CNN Effect” or the “Cockroach Effect”, provides the most appropriate theoretical perspective to understand these multiple, concurrent and indirect forms of media influence. These findings have important implications for government donors, news organisations and aid agencies, and for our wider understanding of how news coverage may influence foreign policy.

financing for appeals related to crises in Zimbabwe (26%), Venezuela (24%) and South Sudan (10%) (FTS 2021). These disparities have hugely significant consequences for many thousands of people. For instance, in April 2020, a funding shortfall forced the World Food Programme to halve its food rations in parts of Yemen, despite growing chronic hunger (Barrington 2020).
It is widely believed that the news media influences the distribution of governments' humanitarian aid-alongside assessments of recipient need and the foreign policy objectives of donor states. Berlemann andThomas (2018, 1026), for example, claim that "media reports on natural disasters have a systematic impact on the amount of provided disaster aid". But although there may be a clear correlation between the amount of news coverage a crisis receives and governments' aid allocations (Cohen, Riffe, and Kim 2021), there is currently "little conclusive evidence" (Eisensee and Strömberg 2007, 693-694) that news coverage influences governments' humanitarian policy. Moreover, questions remain about the causal pathways through which such media influence might operate: whether via the "CNN Effect" (Livingston 1997), the "Cockroach Effect" ( van Belle, Rioux, and Potter 2004) or, as we suggest, via "bureaucratic mediatisation"-defined here as the process by which bureaucracies adapt to the rules, norms and values of the media (Hjarvard 2008). Furthermore, there have been no systematic studies examining these questions from the perspective of the policymakers involved in such decision-making, despite the growing interest in the roles played by bureaucrats within complex systems of governance (e.g., Raudla, Douglas, and Mohr 2021;Thorbjørnsrud, Figenschou, and Ihlen 2014). To address these issues, and to further our understanding of the influence of news coverage on aid policy, we present the findings of 30 in-depth interviews with senior bureaucrats working at 16 of the largest, democratic, humanitarian donor countries.
Our analysis is structured according to two distinct forms of humanitarian aid allocation, as interviewees stressed they were shaped by very different considerations, including news coverage. The first involves governments' "emergency" or "reserve" budgets for humanitarian aid, which can be spent at any time throughout the year, in response to rapidly deteriorating, or sudden-onset crises. The second relates to their annual allocations, which are determined on a yearly cycle, and which usually comprises the majority (70-90 percent) of a donor's humanitarian budget.
Policymakers told us that intense levels of sudden-onset, national news coverage can increase emergency humanitarian aid allocations, even if the level of unmet humanitarian need does not require it. We call this a "sudden-onset national news" (SNN) effect. Policymakers say this happens because such coverage triggers multiple other accountability institutions-including elected officials, public opinion and especially ministers-who then apply pressure on aid bureaucracies. This pressure obliges them to announce additional funding, in order to retain their legitimacy. Such pressure appears particularly pronounced in bureaucracies overseen by publicity-seeking ministers.
By contrast, our interviewees describe their annual aid allocations as largely unaffected by news pressure. This was attributed to an apparent disinterest in these budgets amongst the media, the public and politicians. Importantly, however, the bureaucrats we interviewed often believed that the annual aid allocations of other governments were heavily influenced by news coverage. This perception led some donors to try to compensate by allocating additional funding to under-reported crises in what we call the "forgotten crisis" effect. Thus, even when aid allocations were seemingly insulated from media pressure, news coverage appears to indirectly influence outcomes.
Contrary to prior theorisations of media influence on foreign policy in terms of a singular "effect" on a specific policy, our results indicate that multiple forms of media influence are exerted simultaneously, even within the same bureaucracy, via policymakers' relationships with multiple other actors, both inside and outside of their own government (Lusk 2019). Our findings also suggest that "bureaucratic mediatisation", rather than the "CNN Effect" or the "Cockroach Effect", provides the most appropriate theoretical perspective to understand how news coverage influences humanitarian aid allocations.

Does the News Media Influence Humanitarian Aid Allocations?
Humanitarian law and ethics require humanitarian aid to be distributed on the basis of unmet need (Barnett 2011). Multi-country studies find that humanitarian principles of humanity, neutrality, impartiality, and independence do heavily shape humanitarian aid allocations. According to Annen and Strickland (2017), for example, unmet need among recipients explains a "substantially larger share" of variation in humanitarian aid than factors relating to the donor's national interest. However, donors' domestic considerations, such as their electoral cycles and public perceptions, still play an important role in aid allocations (Drury, Olson, and Van Belle 2005;Annen and Strickland 2017). Foreign policy issues, including the proximity, political alignment, colonial history, language and trade links with recipient countries, are also found to be important factors (Fink and Redaelli 2011).
Unfortunately, research into the influence of news coverage on humanitarian aid is more limited. Studies of aid allocations in the US, UK, France, Japan, and Canada by van Belle, Rioux, and Potter (2004, 138) have demonstrated that, "media coverage and the bureaucratic process of awarding foreign aid are clearly correlated". For example, they find that, "every additional article in Le Monde correlates with an increase of an additional $66,640 in aid" (2004,75). They also repeatedly highlight that it is the volume, rather than the content, of coverage that correlates with aid allocations (although more recent research by Cohen, Riffe, and Kim (2021) challenges this).
While most of van Belle et al.'s (2004, 136) analyses relate specifically to development assistance, they also find evidence of a, "substantial, clear and unmistakable" correlation between US disaster aid allocations and coverage of disasters by the New York Times. Similarly, in their study of US foreign disaster assistance, Drury, Olson, and Van Belle (2005) show that, for each additional news story in the New York Times, an additional halfmillion dollars was allocated to a disaster. They even suggest that, "event severity has less of an impact on the amount allocated than media coverage", stating, "one New York Times article is worth more disaster aid dollars than 1,500 deaths" (Drury, Olson, and Van Belle 2005, 470).
However, neither Drury, Olson, and Van Belle (2005) or van Belle, Rioux, and Potter (2004) are able to fully isolate the effect of the media or disentangle cause and effect. As Eisensee and Strömberg (2007, 693-694) explain, "the problem is that news coverage and policy will be correlated even if news has no effect on policy since news coverage depends on unobserved issue salience and political agendas, both of which directly affect policy". In an attempt to address this problem, Eisensee and Strömberg (2007) analyse US relief to disasters at times when there are competing news stories. They find that, all else being equal, disasters are less likely to receive relief if they occur at the same time as other newsworthy events. Eisensee and Strömberg (2007) argue that the only plausible explanation for this is that relief decisions are at least partially driven by news coverage.
While Eisensee and Strömberg (2007, 3) do appear to have isolated and even begun to quantify the influence of the news media in this case, they are clear that their study, "does not uncover the exact mechanism through which this happens". Furthermore, like much evidence concerning the relationship between news coverage and humanitarian aid, it relates only to the practices of one donor: the United States (Berlemann and Thomas 2018). Moreover, this research only tends to examine the influence of news on "emergency" support for sudden-onset crises, whereas most humanitarian aid is allocated annually, to protracted crises.

How Might the News Media Influence Humanitarian Aid Allocations?
The "CNN Effect" and the "Cockroach Effect" Currently, the two most well-known explanations of how the news media might influence humanitarian foreign policy are the "CNN Effect" and the "Cockroach Effect". The CNN Effect suggests that continuous, real-time international news coverage of distant suffering stimulates public opinion to "do something", which places pressure on democratic governments to intervene on humanitarian grounds, particularly in times of policy uncertainty (Gilboa 2005). In this respect, the CNN Effect conforms to the issueattention cycle of policy influence (Downs 1972, 39), whereby a sudden increase in public awareness of an event is accompanied by "euphoric enthusiasm" about society's ability to solve it.
However, there are numerous, well-documented problems with this theory. First, in an era of "hybrid media systems" (Chadwick 2013), it may be outdated to focus solely on the influence of mainstream news on governments' decision-making (see Cooper 2018). In particular, earlier theory about "the CNN Effect" and related ideas about "public opinion" seem ill-suited to addressing the "salami-slicing" of audiences' attention which tends to accompany the proliferation of mainstream and social media outlets. Second, discussions about "the CNN Effect" are prone to conflating many different kinds of influence, such as agenda-setting and policy acceleration (Livingston 1997). The "CNN Effect" also combines two very different types of policy-making (militarised forms of "humanitarian" intervention and humanitarian aid allocation), which may have different degrees and kinds of susceptibility to media influence (Gilboa 2005). Moreover, such work has tended to focus on governments' allocation of "emergency" aid, neglecting the more routinised decision-making about aid allocation which shapes governments' allocation of the majority of their humanitarian aid budget ( van Belle, Rioux, and Potter 2004).
Third, the "CNN Effect" is notoriously media-centric: failing to account for other causal factors known to influence humanitarian intervention such as geo-strategic interests and logistical issues (Fink and Redaelli 2011). Finally, the "CNN Effect" generally conceptualises "influence" as a simple, unidirectional affair: a perspective undermined by later work which stressed that the news media tended to follow (and legitimise) government policymaking, except in rare circumstances where politicians were themselves unsure about their desired course of action (Livingston 1997). Such work has re-enforced dominant perspectives in political communication about cascading activation (discussed in Lusk 2019), which stress that journalists tend to follow the narrative frames of those in power unless there is elite dissensus (Entman 2003). Although the advent of social media means the dynamics of such cascading have changed, the domination by political elites, we are told, has not (Entman and Usher 2018).
The "Cockroach Effect" offers an alternative approach to conceptualising the potential influence of news coverage on governments' allocation of humanitarian aid, by focusing on the politics which takes place within bureaucracies ( van Belle, Rioux, and Potter 2004). Drawing on the "principal-agent model" of bureaucratic responsiveness, it suggests that politicians, or "principals", utilise a number of "subtle control devices" to ensure that aid bureaucracies, or "agents" adjust to their will, without the need for constant monitoring or supervision (2004). One of these "control devices" is the threat of sanctions, when bureaucracies fail to provide benefits to the principal, or cause them reputational problems. Such sanctions could include dismissing senior managers, cancelling projects, reducing departmental budgets, or even closing departments. Van Belle, Rioux, and Potter (2004, 30) go on to suggest that, like "cockroaches … avoiding getting stomped on", aid bureaucrats try to avoid such sanctions by adjusting their actions pre-emptively, in accordance with the cues which they anticipate the principal will use. Specifically, according to this model, aid bureaucrats are assumed to view news content as an indicator of public demand. They, therefore, match aid allocations with levels of news coverage to avoid politicians sanctioning their department for being out of step with public opinion.
Although it was inspired by comparative research, the Cockroach Effect has not yet been empirically tested. It was, as Joly (2014, 589) explains, "concluded inductively and indirectly from (a) the observation that media influences foreign aid and (b) the assumption that foreign aid decision-making is dominated by (domestic) bureaucracies". As a result, there is, "no clear and direct evidence that media influences foreign aid due to [the cockroach theory of] bureaucratic responsiveness" (2014).

Bureaucratic Mediatisation
A final relevant theoretical perspective-but which has not yet been used to examine the relationship between news coverage and humanitarian aid-is "mediatisation". It refers to the process by which non-media institutions internalise and adapt to the rules, norms, and values of the media (Hjarvard 2008). Rather than conceptualising media influence as an "effect" of media texts, mediatisation seeks to explain how the media, through their existence, formats, content and semi-structural properties, come to reshape and restructure politics (Esser and Strömbäck 2014, 11). Thus, this concept enables us to address the broader consequences of the pervasive spread of media for political and organisational life, such as when effects occur because social actors anticipate how the media will behave (2014).
The concept of mediatisation has been criticised for a lack of conceptual clarity (Schillemans 2012). More detailed, meso-level, institutionalist perspectives on mediatisation-including bureaucratic mediatisation-have been developed to address this problem. These approaches conceptualise the news media as a semi-autonomous social institution with its own "logic of appropriateness", or implicit journalistic norms and values. They examine the extent to which political institutions pursue legitimacy by adapting to a "media logic" rather than a "political logic" (Esser and Strömbäck 2014, 15). In the case of foreign policy decision-making, for example, Brommesson and Ekengren (2017, 26) argue that when states adopt a media logic they respond to, "media demands that currently dominate the polarized public debate" by pursuing policies that are "believed to reflect the popular will of the majority".
Bureaucratic mediatisation focusses specifically on how the unique organisational structure and behaviour of rule-based public organisations are affected when they adopt or adapt to media logics (as opposed to, for example, political logics) (Schillemans 2012). Thorbjørnsrud, Figenschou, and Ihlen (2014, 14) have identified several news logics with particular significance for bureaucracies. These include a preference for personalisation and simplification but also a "news rhythm" or, "requirement to be first or on time with a breaking story".
The concept of "media pressure" is central in these analyses, explaining why government agencies might act according to a media logic (Schillemans 2012). According to the "media pressure" hypothesis, "government agencies that are more regularly scrutinized, portrayed, and criticized by the media can be expected to have stronger incentives to adapt to the media" (Fredriksson, Schillemans, andPallas 2015, 1053). The assumption is, as Jacobs and Schillemans (2016, 30) explain, that by reporting on bureaucracies, "media can be sparks for accountability because the perceived risk of negative publicity invokes anticipatory reactions in public organisations". Such pre-emptive self-criticism is, they suggest, most likely to be linked either to sudden news coverage of a relevant "focusing event" (Birkland 1998), or to critical coverage of their organisation. However, unlike the "Cockroach Effect", bureaucrats "blame avoidance" practices are thought to be more likely to involve strategic communications and relationship-building with journalists, rather than the re-allocation of resources (Schillemans 2012).
News coverage may also place pressure on bureaucracies indirectly, by triggering other accountability institutions that monitor governments, such as civil society organisations, elected officials and the public. Jacobs and Schillemans (2016, 14) find that in early 2010, more than two-thirds of parliamentary questions in the Netherlands were prompted by news reports. For this reason, they describe the news media as the, "prime 'connecting mechanism' between the different entities monitoring governments" (2016). Yet media pressure can also be triggered by these other accountability institutions when the news reports on their actions. They conclude that "the media, thus, play an integrative role in fragmented processes of accountability" (2016).
The effects of news pressure on public bureaucracies have also been shown to depend heavily on the actions of the government minister(s), or Secretaries of State, overseeing the department. For example, based on a study of Norwegian government ministries, Figenschou et al. (2019, 377) conclude that "media pressure alone is insufficient" to change policy decisions unless it coincides with the interests of the minister. Specifically, they argue that "for politicians, media attention offers an opportunity to demonstrate agency, implement policy and send a signal to political opponents and/or the citizens" (Figenschou et al. 2019, 389).
In this respect, ministers engage in what Esser and Strömbäck (2014, 15) describe as a "frontstage" publicity-gaining political logic, which focusses on, "tactics and strategies for winning public support and publicity". However, a minister's potential role in proactively, "using media pressures to tilt decision-making towards a specific direction" (Figenschou et al. 2019, 379), raises the question of whether bureaucratic adaptation to a news logic is a result of news pressure or political pressure. In response, Figenschou and her colleagues (2019, 381) suggest we think of media and political pressures as operating simultaneously in an "unpredictable" and "dynamic ebb and flow".
Research into bureaucratic mediatisation has been criticised for depicting bureaucracies as, "more or less-unreflective and passive victims of mediatization" (Fredriksson, Schillemans, andPallas 2015, 1053). Yet in their study of Swedish bureaucracies, Fredriksson, Schillemans, andPallas (2015, 1049) found that they, "have substantial agency in terms of how they cope with the media". Moreover, they conclude that bureaucracies', "propensity to adapt to the media is mainly determined by their management structure rather than, as could have been expected, by media pressure" (2015). Specifically, organisations overseen by career managers tend to invest more in media management than those led by policy specialists. This is because the latter take their professional peers as their most important points of reference and so prioritise the norms and values of their specialist field above the norms of their organization (2015).
Bureaucratic responses to news media are also shaped by traditional bureaucratic logics, such as impartiality, neutrality, correctness and equal treatment, which form the basis of bureaucratic legitimacy. In an ethnographic study of the Norwegian Ministry of Justice, Thorbjørnsrud (2015) found that these public service norms led bureaucratic responses to media pressure to be shaped by an "administrative loyalty" to the bureaucratic system, and not just by an imperative to serve the interests of political superiors. Moreover, in a comparative study of European countries, Raudla, Douglas, and Mohr (2021) found there was a correlation between bureaucrats' autonomy and their tendency to take a more technocratic, depoliticised approach to problem-solving.
In summary, within the study of bureaucratic mediatisation, the influence of the news media is understood to operate within a complex and dynamic system of governance, involving interactions with multiple other accountability institutions, alongside various other political, professional and bureaucratic logics. Despite their significant theoretical value, however, empirical studies of bureaucratic mediatisation are, "scarce, scattered and hard to come by" (Schillemans and Pierre 2016, 2). In particular, "mediatization research and foreign policy analysis are still highly unknown to one another" (Brommesson and Ekengren 2017, 190). We therefore aim to contribute to this literature by testing the applicability of bureaucratic mediatisation to humanitarian aid bureaucracies, by asking: RQ1: Do policy makers believe that news coverage influences the allocation of official humanitarian aid? If so, how do they believe this influence operates? RQ2: Which theoretical approaches are most helpful in enabling us to understand bureaucrats' perceptions of the influence of the news media on humanitarian aid?

Methods
We examine the perceptions of decision-makers at 16 of the largest, democratic, humanitarian donors, using semi-structured interviews. These were: Australia, Belgium, Canada, Denmark, the EU Commission, Germany, Ireland, Japan, New Zealand, the Netherlands, Norway, Spain, Sweden, Switzerland, the UK, and the USA. In 2020, these donor countries were collectively responsible for over 90 percent of all humanitarian funding for UN response plans/appeals (FTS 2021). Non-democratic donors were excluded as the political, bureaucratic and media logics involved are likely to be very different.
We conducted 30 in-depth, semi-structured interviews with senior civil servants responsible for humanitarian aid allocation decisions within these donors. This included 12 departmental directors, or equivalent, 5 deputy directors, 7 senior advisors, 2 communication directors, 2 desk officers and 2 senior diplomatic representatives to multilateral organisations. Seventeen of our interviewees were female and thirteen were male. In most cases, humanitarian departments were relatively small, meaning that, as one interviewee put it, "each one of us has a lot of influence about the decisions we take". Thus, while we only interviewed, on average, around two individuals per donor, in each case they had substantial decision-making authority and represented a significant proportion of the population of relevant decision-makers.
Gaining research access to senior bureaucrats is challenging. Indeed, this study is unique in having achieved access to so many different ministries. We are indebted to the UN Office for the Coordination of Humanitarian Affairs (UN-OCHA) for their assistance in helping us to identify relevant interviewees, whilst respecting our scholarly independence and placing respondents under no obligation to participate. All interviews were carried out between April 2019 and September 2020, either face-to-face or remotely, and lasted between 30 and 75 minutes. Interview transcripts were coded using NVIVO and analysed thematically. Data analysis was conducted by the lead author to ensure consistency, and regular team meetings were held to discuss emerging themes.
A key limitation of interviewing is that respondents often wish to represent themselves in a positive light (Robson and McCartan 2016). This seems especially likely if, as the literature reviewed above suggests, bureaucrats seek to avoid negative publicity, internal criticism or other sanctions. Indeed, almost all interviewees regarded media influence over their decision-making as undesirable because it contravened humanitarian principles. One interviewee even told us that it would be "career limiting" for them to be on record saying "I think the government responds to media in how it decides [its humanitarian aid]".
To mitigate risks to interviewees' livelihoods, we assured all respondents that they were participating on the basis of full anonymity. This anonymisation prevented us from disaggregating our results by country or highlighting country-specific phenomena. We also used sensitive interviewing techniques to encourage participants to "open up" (Robson and McCartan 2016, 286). Before asking interviewees directly about media influence, we posed a series of more open, exploratory questions about related issues, including their perceptions of "public opinion", departmental sanctions, ministerial influence, strategic communications, non-media sources of pressure and professional and bureaucratic logics.
In order to triangulate our interview data, we also analysed all publicly available documents relating to the 16 donors in our sample, which described the processes and/or criteria used for determining humanitarian aid allocations. This included policy and strategy papers, annual reports and some departmental evaluations. In total, 28 documents were identified, with at least one relevant document for each donor. In all cases, the principle inputs, logics, and outputs relating to decision-making over humanitarian aid allocations were established. In addition, documents were examined for any explicit or implicit references to the news media, strategic communications or public opinion.

Emergency Aid Allocations
Over half of our interviewees stated that intense news coverage of a humanitarian crisis had, in certain circumstances, created pressure on their department, which led to an increase in emergency aid allocations. In perhaps the clearest illustration of how news coverage created such pressure, and how it could affect aid allocations, one interviewee explained that, Where the media will have an impact is if something is suddenly breaking out. Then various feedback loops will get back to me and the team at a technical level, to think, 'should we be doing something about this?' More importantly, it could get to ministers and say, 'Why aren't we responding to this cyclone?' It will get back to the public as well; we will start getting letters from the public, answering parliamentary questions. So, there are multiple ways in the system where, if something becomes a big media issue, all of a sudden, right through the system, people just start asking questions about it … So, there [is] a real pressure to be able to say, 'this is how we responded'.
Interviewer: Because of the news coverage specifically?
Yes. Because, in terms of the number of people affected and the need for international involvement, it's actually pretty low. But we still needed to be able to say, 'This is what we have done.' … The political imperative is to have funding within 24 or 48 hours.
Such instances of media influence were mentioned in relation to a relatively narrow range of specific contexts, including the ongoing crisis in Yemen, the 2020 Beirut explosion and the 2015 Rohingya refugee crisis. A further four interviewees, who claimed to have never been influenced by the news media in this way, suggested nevertheless that such influence was plausible. One such respondent said, "You could foresee a scenario where there is a sudden crisis that is top of the media and … you have to do something quickly". Only three out of 30 interviewees claimed that such media influence was impossible, with one commenting that, "This non-interference is, I am fully aware, a privilege". Instead, they claimed that their aid allocations were based entirely on assessments of humanitarian need.
What Kind of News Coverage Leads to Pressure on Aid Bureaucracies? According to our interviewees, there is a direct correlation between the intensity of news coverage, and the level of pressure exerted on emergency aid allocations. As one respondent said, "the bigger the coverage … obviously, there is going to be a stronger political uptake for it". In line with the "media pressure hypothesis" (Fredriksson, Schillemans, and Pallas 2015), interviewees repeatedly described such intensity as relating to the volume and prominence of news coverage, rather than its content (Cohen, Riffe, and Kim 2021). As one interviewee told us, "the rule of thumb was … if it was the number one item on the evening news … then it's a big crisis [and] we need to be able to at least say, 'we have done this'". Similarly, in the UK's Guidance Framework for Humanitarian Response in Sudden Onset Contexts (CHASE 2017), the "level of media interest" refers to the extent to which "stories are appearing in multiple segments of the mainstream UK media".
In addition, interviewees repeatedly linked media pressure with the "news rhythm" (Thorbjørnsrud, Figenschou, and Ihlen 2014), or the abruptness or sudden-onset of coverage. This was evident in multiple references to "sudden crises" and a "surge in media interest". One interviewee also said that "the more intense pressure comes around the natural disasters … [Because] there is a lot of interest and a lot of coverage … In the case of natural disasters, I would say definitely media coverage just 100% drives pressure".
However, not all instances of news pressure related to intense news coverage of disasters or rapid deteriorations in protracted crises. Several interviewees described how the UN's "four famines" campaign in 2017 "forced us … to develop some allocations for those countries [even though] most of them were not covered by our priority contexts". This campaign referred to four separate crises related to food insecurity-in Sudan, Nigeria, Somalia and Yemen. Despite its name, only one region in one of these four countries was experiencing a famine at the time (Unity State, South Sudan), and none were sudden-onset crises. However, the collective framing of these crises as "four famines" generated a sudden and relatively large volume of news coverage which, according to several interviewees, "helped mobilise funding". This case is significant for several reasons. Most obviously, it demonstrates that it is sudden-onset news coverage which activates pressure, rather than sudden-onset crises per se. In addition, it shows that news pressure can be activated by aid agencies, as well as journalists. But it also illustrates the complex relationship between news values and the media pressure experienced by bureaucrats. Aid bureaucrats did not perceive these food crises in the same way as journalists, who rely on news values such as unambiguity, continuity, timeliness and negativity to help determine what they cover (Harcup and O'Neill 2017). Yet, as other scholars have found (Garland, Tambini, and Couldry 2018;Thorbjørnsrud, Figenschou, and Ihlen 2014), the operation of a "news logic" within bureaucracies meant that, in certain circumstances, many bureaucrats were compelled to respond as if they shared journalists' value-judgements.
Interviewees consistently said that the most important news sources were national media-especially television, newspapers and radio. This is because such "mainstream media" (CHASE 2017) were still perceived as, "the biggest influencers of decisions and opinions in [the country]". As one interviewee explained, "national media [matters] … not specifically to me, but for our politicians. They are elected in [this country], so … usually what matters is their publicity in [this country], how popular they are in [this country] … It's media that reaches the larger public".
Despite the emphasis on international news outlets, and especially satellite TV coverage, within "the CNN Effect" (Gilboa 2005) international news organisations were only described as significant if they influenced domestic coverage. As a representative of one non-European donor said, "In terms of what drives voters, it is not reading The Guardian. That doesn't get anybody elected. But, that being said, when big crises get big media coverage internationally, that will sometimes get picked up [by national media]".
Finally, despite growing critical interest in "hybrid media systems" (Chadwick 2013), and the importance of social media to bureaucrats in other studies (Garland, Tambini, and Couldry 2018), our respondents seemed much less interested in, and responsive to, social media. Very few said they spent much time monitoring social media, and the small number who did tended to discuss social media in purely instrumentalist terms: as a means of identifying relevant articles by specialist news outlets and bloggers. Even these bureaucrats stressed that social media was not a significant factor in shaping their decision-making, as they argued that it did not have a sufficiently, "wide reach" or "cut through to the public or politicians".

How Does News Coverage Create Pressure on Aid Bureaucracies?
In line with previous studies of bureaucratic mediatisation, our interviewees said that news influence on aid bureaucracies operates largely indirectly-by "triggering interest", as one interviewee put it, from multiple other accountability institutions simultaneously. This is illustrated by the references to "various feedback loops" and "people asking questions … right through the system" in the earlier quotation. Similarly, another respondent told us that, "media coverage … drives pressure, not only from elected officials, but from all parts of government, from our leadership here in the [aid department], to leadership in parliament and in the [ministry of foreign affairs]". These various accountability institutions applied pressure directly on the aid bureaucracy by, "calling us and asking us a lot of questions about what we are doing".
Members of the public could also exert pressure on aid bureaucracies, in response to news coverage, by sending enquiries to their governments. Notably, some aid bureaucracies regarded levels of news coverage as a proxy measure of public opinion, in these circumstances. For example, in New Zealand's Humanitarian Action Policy (MFAT 2019), "media interest lasting more than 48 hours" is one of two indicators used to assess where there is "significant and sustained New Zealand public interest" in a "large-scale rapid onset emergency".
Respondents also described how news pressure tended to operate largely through ministers. For instance, one interviewee explained that "Media have an influence … on our parliament, on government, on the opposition. So, when everybody starts [putting] the pressure on our politicians, they put the pressure on us". For this reason, there was a clear consensus amongst interviewees that pressure exerted by ministers was by far the most significant. For example, when asked whether they were particularly sensitive to interest from any one source, one respondent replied, "Obviously, there is a seniority question: if the foreign minister sends you an email saying, 'Why haven't you responded to this?', you need to". Furthermore, respondents described how some ministers reacted to sudden-onset national news coverage by proactively using it as an opportunity to engage in a "frontstage" publicity-gaining politics (Esser and Strömbäck 2014). As one interviewee put it, Within a couple of hours after the earthquake, our minister already announced her contribution. It's not even clear where it is going to, but she just wants to be in the press release [saying], "Our minister is giving so much money to the victims".
Bureaucrats believed that the purpose of such publicity was not to pre-empt critical external scrutiny of governments, but to advance the personal career/s of the minister/s concerned. This is made clear in the following quotation.
Sometimes the media report a lot about [a crisis] and the minister will send us messages, [saying] 'we should do something' … Their priorities are not purely humanitarian. It is also about their career … To guarantee their next steps.
[Some] ministers like to [have] publicity … Then the public will like them … We see that especially with ministers that don't specifically have a humanitarian career. Therefore, the personal priorities of the minister-and the degree of autonomy an aid department has from them-appear key to explaining variations in bureaucratic mediatisation between donors. As one interviewee explained, "It's to do with the minister's personality … This minister is not so sensitive to newspapers … [but] some are more receptive to the media". Indeed, several interviewees described a change of minister as leading directly to a significant shift in the degree of news pressure exerted upon them.

How Does News Pressure Affect Emergency Aid Allocations?
News pressure appeared to influence emergency aid allocations by pressuring aid bureaucracies into temporarily basing their decisions on a particular kind of media logic, involving considerations about how policy decisions would "play" in national news media (Esser and Strömbäck 2014). Adopting this sort of media logic requires aid bureaucracies to take, "rapid political decisions to stop the ongoing catastrophe" (Brommesson and Ekengren 2017, 5) by announcing relatively large, additional emergency aid allocations-even if this was disproportionate to the actual level of unmet need. This is illustrated in the earlier quotation by the references to, "a real pressure to be able to say, 'This is how we responded, [even if] the number of people affected … [is] actually pretty low".
Similarly, numerous other interviewees described news pressure as creating an "obligation" or "imperative" to "have something to communicate". One respondent spoke of "a direct pressure to make sure that we are doing something so that we can talk about it". This apparent "obligation" to adapt to a media logic occurred because it was perceived to be the most effective way of gaining or retaining legitimacy for aid allocation-decisions, in these exceptional circumstances. As one interviewee said, "You must have symbols of solidarity … You must say, "We will give enough money to have clear links [to the crisis]" … To make this very, very visible … [Otherwise] you will lose your legitimacy".
Almost all interviewees regarded this adoption of a media logic as, "a failure of our mandate" or "very wrong in terms of humanitarian principles". For example, one respondent explained that this news pressure, " … works, but I don't know if it is very humanitarian". Specifically, they felt that conventional news values, such as cultural proximity, led news coverage to privilege some crises over others, in ways that are wholly unconnected to levels of humanitarian need. Indeed, previous research has repeatedly shown that a small number of "high profile" humanitarian crises receive the vast majority of news coverage, leaving others marginalised and hidden (Hawkins 2011, 59). Such disparities have repeatedly been shown to correspond primarily with the geopolitical significance of a crisis and its cultural proximity to an audience, rather than with its severity, or the number of people affected (see Joye 2010).
For this reason, interviewees frequently sought to "anticipate" and resist media pressure. One of the most successful strategies in these circumstances was to contribute, in advance, to pooled, flexible funds, such as the UN's Central Emergency Response Fund (CERF). Interviewees described this kind of needs-based, rapid-response funding as enabling them to, " … immediately say to our minister, 'Yes, our money is contributing. So, we don't have to do anything additional there".
However, such pre-emptive resistance strategies were not always successful. The "news rhythm" associated with sudden-onset reporting required aid bureaucracies to make decisions about emergency aid allocation very rapidly. This requirement to, as one respondent said, "do something quickly", is highlighted in the earlier quotation by the reference to a "political imperative … to have funding within 24 hours or 48 hours". Similarly, another respondent said that, "Emergency funding is happening very fast, and we need to find fast arguments because we need to make fast decisions".
This speed is significant because it dramatically reduced the amount of time bureaucrats had to anticipate and respond to their ministers' concerns. It also gave bureaucrats little time to gather information from seemingly objective or expert sources, which were ordinarily used to inform, strengthen or defend needs-based, rather than news-led, decisions (a strategy that was used to protect annual aid budgets, discussed below). As a result, the legitimacy ordinarily provided by humanitarian logics was undermined and the pressure stemming from intense, sudden-onset news coverage became more difficult to resist. It is also worth noting that not all respondents regarded media influence as inherently undesirable. A minority felt that reacting to media coverage was "democratically appealing" (Figenschou et al. 2019, 381) because it indicated that the government was seeking to be responsive to public opinion. One also felt that media influence could potentially help correct some governments' tendencies to allocate aid based largely on national self-interest, rather than humanitarian need.

Annual Aid Allocations
In marked contrast to their discussion of emergency aid allocation, interviewees stated that news coverage did not influence annual allocations of humanitarian aid. They gave three reasons for this. First, in virtually all cases, there was a widespread perception within aid bureaucracies that their annual aid allocation decisions received very little media attention. As one interviewee told us, "In [this country] there is really not much media attention on … humanitarian issues. There is really, really not … In terms of '[our] aid budget did x or y', it is virtually nil". Interviewees also repeatedly described annual humanitarian aid as receiving very little critical media scrutiny, particularly in comparison with coverage of international development spending. For example, one told us that, "Even our media, I think, is less critical of humanitarian [spending] than it is of the development [budget] … So, it is a bit protected from that".
According to the "media pressure hypothesis" (Fredriksson, Schillemans, and Pallas 2015), this lack of media attention-critical or otherwise-gives aid bureaucracies little incentive to adapt to a media logic: because other accountability institutions are less likely to be prompted to scrutinise their decisions. This was confirmed by the testimonies of multiple interviewees, who described the media as " … not … the pressure makers". For example, one interviewee told us that, when determining their annual aid allocations, "[There] can still be pressure … driven by an [elected official] who has a pet issue, or by colleagues in the building, who think their country is not receiving enough attention … but it is just not often directly driven by media".
Second, the apparent lack of news pressure on annual aid allocations stemmed from the widespread perception amongst bureaucrats that, ordinarily, the public were either uninterested in, or already broadly supportive of, annual aid allocations. As one interviewee said, "I don't think people are very interested outside of natural disasters". For this reason, bureaucrats did not consider it important to monitor levels of public support via the news media or elsewhere. There were no references to opinion polls or any other inputs relating to public opinion when determining annual aid allocations in any of the policy documents examined. In a rare exception, one interviewee did say that "I think politicians listen to the public … I think they are conscious of the public mood". In general, though, interviewees did not believe that levels of news coverage influenced, or acted as a proxy for public opinion, outside of sudden-onset reporting.
Third, bureaucrats told us that the longer lead-in time involved in annual aid allocations meant that they were able to respond to ministers' "frontstage" publicityrelated concerns via planned, strategic communications, rather than by adjusting their annual aid allocations. Despite the widespread perception that both the public and the media were largely uninterested in annual aid allocations, and the time-consuming nature of providing information to their communication departments, a publicityseeking politics did still operate within most aid bureaucracies. Indeed, several interviewees described "giv[ing] more publicity to what we are doing" to "get more support from the voting public", as part of a "political game". One said, "It's very important to communicate about … our humanitarian funding. Because every time we pay money to something … the minister would like to get credit for it". But bureaucrats also stressed that this kind of publicity-seeking did not "affect our decision[s]". In fact, one interviewee even described their strategic communications as a kind of defence against ministerial pressure, stating, "If you don't make sure that what you propose has a broad support in society, then you end up doing what the minister thinks this morning, after having breakfast and hearing the news".
As a result of the above factors, bureaucrats claimed that ministerial interference with annual aid allocations was extremely rare-whether prompted by news coverage or any other feedback loop. As one respondent told us, "I have been here with three different directors of the agency and three different vice-ministers, and they have not changed a comma in what we proposed".

Resisting News Pressure
In the rare circumstances where bureaucrats did feel under pressure to modify their annual aid allocations, whether because of news pressure or other influences, they claimed they were able to anticipate and resist such pressure, in almost all cases. Such resistance generally involved providing further evidence or explanations, usually in the form of quantitative, indicator-based analysis, to justify their original decisions. Many bureaucrats also pro-actively engaged with the relevant minister and other parts of government, to identify and respond to any concerns, before submitting their recommendations for approval. As one interviewee explained, "There's a lot of ongoing work, not just with the minister's office, but with colleagues in development, foreign affairs … [and] embassies. It's a constant maintaining of relationships to make sure everybody understands our work and we listen to their input".
Bureaucrats' "relationship management" practices did not seem to be motivated by their concern to protect or promote their careers, as the "Cockroach" theory suggests. At least, such concerns were never mentioned by interviewees. Instead, study participants claimed that they were driven, as Fredriksson, Schillemans, and Pallas (2015) suggest, by their adherence to professional, humanitarian norms associated humanity, neutrality, independence and impartiality (Barnett 2011). These dictate that aid allocations should be governed by levels of unmet need alone. As one interviewee told us, "We try to be guided strictly by the humanitarian principles, by the severity of needs, which is part of the independence and impartiality of the principles".
Interviewees also argued that this adherence to conventional humanitarian norms helped provide them with sufficient legitimacy to successfully defend their annual aid allocation decisions. They frequently described such norms as providing "credibility" or as helping them to "make a strong case" or tell "a good story" to ministers. As one respondent explained, "The minister has a lot of opinions [on] a lot of stuff we do, but because it is based on the humanitarian principles, he approves of it … It is hard to argue against humanitarian principles, because those are absolute". Similarly, several interviewees and documents described the principles of "Good Humanitarian Donorship" (GHD) as providing an effective way of preventing "political influence". One respondent told us, "Parliament may be responsive to media and we are certainly responsive to our parliamentbut we are principled donors and that helps protect our independence". Moreover, since annual aid allocation decisions were not made under severe time pressure, bureaucrats were able to offer more comprehensive, comparative evidence to support their recommendations. This further enhanced their legitimacy.
The "Forgotten Crisis" Effect In the absence of significant news pressure, interviewees and policy documents repeatedly emphasised that annual aid allocations were governed by a combination of factors relating to recipient-need and donor-interest, as previous research has suggested (Annen and Strickland 2017). News coverage was widely perceived as an unreliable way of gauging either issue, especially national news coverage, which a significant majority of participants regarded as being of "low quality" with a lack of "accuracy", "understanding" and "context". This mirrors the findings of previous studies, which have repeatedly shown that mainstream news coverage of humanitarian affairs is often sporadic, simplistic and de-contextualised (see Scott, Wright, and Bunce 2022).
Despite this, we found that news coverage did still play a role in most donors' annual aid allocation decisions-although not in a way anticipated by previous research. In addition to supporting crises with the greatest unmet need, a majority of donors allocated annual aid to crises that they judged to be "neglected" or "forgotten". For example, Ireland's humanitarian assistance is described as in one document being, "directed to where needs are greatest, with particular emphasis on targeting forgotten and silent emergencies". The European Commission's department for overseas humanitarian aid (ECHO) even earmarks 15 percent of its funding to such "forgotten" crises. Interviewees explained that this stemmed from a concern for "adding the most value" or "making the most difference" with their funding. According to Krause (2014), the concept of "adding value" has become a conventional humanitarian norm because it accommodates circumstances in which resource constraints or logistical challenges prevent aid agencies from providing relief based entirely on levels of unmet need.
In almost all most cases, interviewees stated that the news media played a key causal role in their assumptions about whether a crisis had been "forgotten", or would have a relatively large funding shortfalls compared to other crises. This belief is reflected in related policy documents. For example, ECHO (2011, 5) describes the "insufficient international aid" given to "forgotten crises" as being "due in part to a lack of media interest". It also directly informs the design of ECHO's "Forgotten Crisis Assessment" (FCA) Indexthe principal input used by almost all donors to determine which crises are "forgotten". The FCA Index includes "levels of media coverage" as one of its four, equally weighted, criteria for establishing a crisis as "forgotten" and therefore often qualifying for additional funding.
Underpinning these kinds of judgements is an assumption that the humanitarian aid allocated by other governments is determined partly by levels of news coverage. As one interviewee put it, "Generally, if they are not covered by the media, it means that, most probably, other donors will not fund this crisis". Such susceptibility to media influence was perceived by almost all interviewees to be highly undesirable because it contradicted humanitarian principles of prioritising unmet needs. Bureaucrats said they sought to compensate for this perceived distortion in other governments' funding practices by allocating more humanitarian aid to crises which did not receive extensive news coverage. In doing so, these respondents understood themselves to be acting according to a humanitarian logic. As one interviewee told us, "we consider that we have an added value in financing these forgotten crises". Thus, although governments' annual aid allocations do not appear to be influenced by news pressure, a belief within aid bureaucracies that they are, leads a majority of donors to interpret a lack of news coverage as an indicator that greater support is needed.

Conclusion
Most of the policymakers we interviewed told us that sudden-onset, national news coverage of humanitarian crises creates a pressure on aid bureaucracies that can lead to an increase in emergency humanitarian aid allocations to specific crises. This is a significant finding, given that almost all interviewees regarded media influence over aid allocation as undesirable, and therefore may have been tempted to understate, rather than overstate, its influence on their work. Furthermore, interviewees said that national news had the greatest influence-rather than international news coverage or social media-because such coverage was seen to reflect and inform national "public opinion". Our study also suggests that this apparent "Sudden-onset National News" (SNN) effect has more influence on bureaucracies with a relative lack of autonomy, overseen by publicityseeking ministers.
However, interviewees also claimed that annual aid allocations-which make up a much greater proportion of humanitarian aid budgets-were far less susceptible to news pressure. This is significant because it suggests that the news media's influence over aid bureaucracies varies greatly between different forms of aid allocation. This is an important finding since most previous studies of media influence over humanitarian aid allocation focus only on "emergency" support for sudden-onset crises.
Our analysis also sheds light on how policy-makers understand media influence on emergency humanitarian aid to operate: by triggering multiple other accountability institutions, to apply pressure on aid bureaucracies to temporarily adapt their decisionmaking to a media logic. This corresponds with the findings of previous studies of bureaucratic mediatisation (Figenschou et al. 2019) and contradicts many of the assumptions within the CNN Effect and the Cockroach Effect. In contrast to the CNN Effect, it suggests that the international media may be much less influential than previously thought. In addition, influence on emergency aid occurs, not simply because of either news-driven public pressure or direct news pressure on elites, but via a complex combination of both, in tandem with pressure from multiple other accountability institutions. The "four famines" case (discussed in What kind of news coverage leads to pressure on aid bureaucracies?) also demonstrates that such news pressure can itself be triggered by these other actors.
Our interviewees also suggest that news pressure is triggered, not by continuous, realtime coverage of humanitarian crises, but due to the intensity and sudden-onset nature of such coverage, combined with the opportunity it provides for politicians to gain favourable publicity. Finally, bureaucrats do not believe that elite dissensus or policy uncertainty makes media influence more likely, in these circumstances. In fact, bureaucrats were generally very aware of the response required to meet levels of unmet need, even for suddenonset crises. Rather, a media logic was adopted because it temporarily provided greater legitimacy than humanitarian norms-the legitimacy of which is temporarily undermined by time pressures created by the rapid "rhythm" of news media.
In contrast to the Cockroach Effect, our findings suggest that, when determining annual aid allocations, bureaucrats seldom rely on news coverage as a reliable indicator of public opinion. Moreover, they do not appear to act primarily according to self-interest, by seeking to avoid both publicity and ministerial attention. Instead, our results highlight the significance of bureaucratic agency and humanitarian norms. These norms help explain why our respondents sought to resist, rather than respond to, both news pressure and publicity-seeking pressure from their ministers, when possible. For this reason, we conclude that civil servants working in aid bureaucracies are analogous, less to cockroaches, and more to humanitarians. However, as with all the inferences we draw from this study, this conclusion may have been shaped by our reliance on bureaucrats' selfreporting.
Yet bureaucrats' perceptions matter-even if they might be fallible. Many respondents believed that news pressure routinely affected the annual aid allocations of other governments, but not their own. In tandem with the humanitarian principle of "adding value", this belief led many respondents to interpret a relative lack of news coverage as grounds for increasing their annual aid allocations, in what we characterise as a "forgotten crisis" effect. As a result, we suggest that the news media's greatest influence over annual aid allocations may occur, not as a result of intense news coverage or the temporary adoption of a media logic but, ironically, due to a humanitarian logic and a lack of news coverage. This contradicts "the other side of the CNN Effect", or Hawkins's (2011) suggestion that a lack of news coverage of "stealth conflicts" contributes to lack of policy.
Given that governments' annual aid allocations may not be responsive to news coverage in the way that many bureaucrats assume, it follows that assessments of "forgotten crises", such as the FCA Index, may need to be amended to reflect this. Other implications for donors seeking to mitigate or avoid the potential influence of the news media include; providing more support to flexible funds like CERF, supporting the international Good Humanitarian Donorship (GHD) Initiative, building public understanding of humanitarian principles and developing more transparent, evidence-based allocation methodologies to help defend aid allocations. For actors seeking to exert influence on donors via the news media, these findings suggest they should target national news outlets during periods of intense, sudden-onset coverage (or seek to trigger such coverage themselves). They should also target governments with less autonomous aid bureaucracies, career-oriented ministers and larger emergency aid budgets.
In relation to our wider understanding of the influence of news coverage over foreign policy-making, these results suggest moving away from theorising about a single "media effect" on a specific policy-to considering multiple influences, operating simultaneously, within the same bureaucracy. In the case of humanitarian aid allocations, the most significant appears to be a "forgotten crisis" effect and a "SNN effect". Moreover, these influences should be understood, not as exerted directly by the news media on bureaucracies, but as prompted by the interaction of a wide range of actors, whose responses are shaped by their own organisational and professional logics.