18.1 Introduction
Welfare economic theory seeks justification for government intervention in markets in the context of market failure and distributional issues.Footnote 1 In the absence of market failures, free market forces will lead to maximum welfare. This may, however, render a distribution of welfare that insufficiently reflects societal preferences. This situation can then require further redistributive intervention. An analysis of the market failures that exist in a specific industry or market can not only provide justification for government regulation or other kinds of intervention in general but also suggest which type of intervention or regulation is optimal from a welfare economic perspective.
For better or for worse, the creation, distribution, and consumption of news – be it in newspapers, on radio or television, or online – is subject to many different forms of regulation and intervention in almost every jurisdiction and throughout time. Interventions range from censorship to subsidisation and from protection of journalists to specific rules on market concentration and ownership transparency.Footnote 2 While many such rules do not stem from a welfare economic analysis, several market failures are at play in news markets. This chapter focuses specifically on the ‘public good’ character of news and the positive externalities associated with news production and consumption. The market failures in this arena constitute the core problem underlying private investment in news production. The question that this chapter addresses is how these market failures have been affected by the emergence of news aggregator platforms on the one hand, and the introduction of generative AI in news production on the other. In this chapter, a news aggregator platform is defined as ‘an online platform or software device that collects news stories and other information as that information is published and organizes the information in a specific manner’.Footnote 3 As Dellarocas and colleagues describe: ‘[c]ontent aggregators produce little or no original content; they usually provide titles and excerpts (hereafter called snippets) of the articles they link to. Examples of well-known aggregators include Google News, the Drudge Report, and the Huffington Post.’Footnote 4 Such aggregators have gained an increasingly important role in present-day news consumption.
The role of generative AI in news production is twofold. First, it concerns the use of generative AI tools by traditional news media to produce news articles from ‘raw data’ such as data on sports matches, weather forecasts, or stock market data. This will be referred to as AI journalism in this chapter. The replacement of human journalists by AI journalism, which is both ongoing and a development very much feared by many, is a timely issue and also the core topic of Chapter 21 by Martin Senftleben in this volume.
Second, generative AI can be used by companies outside the traditional news media to create news articles based on other news articles and information that the AI system has access to. By doing so, such AI systems serve as a new generation of aggregators, both producing and not producing their own content, as will be fleshed out in this chapter. This output will be referred to as AI news digests in this chapter.
The focus of the analysis is on the public good character of news and the positive externalities of news production. The question addressed in this chapter is: Have the consequences of these existing market failures become more prominent or have they been resolved by these developments? Based on this analysis, the chapter discusses how this informs public policy concerning these developments. By doing so, this chapter takes a welfare economic perspective, which for reasons of scope is necessarily a bird’s-eye view.Footnote 5
This chapter is structured as follows. Section 18.2 discusses the public good character of news and the positive externalities associated with news consumption, and analyses the implications these have for investment in news production. Section 18.3 analyses how the growing role of news aggregator platforms and generative AI affects factors. The concluding Section 18.4 discusses potential remedies.
18.2 Externalities and the Public Good Character of News
News, whether published in a newspaper, on the radio, on television, or online, is primarily an information good. As such, it shares many of the economic characteristics of other information goods or data. Shapiro and Varian pointed out in their bestseller Information Rules that ‘[i]nformation is costly to produce but cheap to reproduce’.Footnote 6 In economic terms, information is characterised by relatively high fixed costs of production and low marginal costs of reproduction, or even (practically) zero marginal costs in the case of digital information. This implies that information goods need to be priced based on customer value – tailored to customers’ price sensitivity through specific discounts or premiums – rather than marginal or even average production costs, in order to recoup the fixed production costs.
18.2.1 News Facts as a Public Good
Two other important features of information are that by nature, information is non-excludable and non-rival.Footnote 7 The non-rival nature of information means that the consumption or use of a specific piece of information by one person or company does not affect the use by another: it cannot be ‘used up’. If a person watches or reads the news, it does not prevent others from watching or reading the same news.
Non-excludability means that access to (published) information via its copying and distribution cannot be controlled. Together, non-excludability and non-rivalry define a public good.Footnote 8 Private provision of public goods is usually problematic, primarily as a result of their non-excludable nature. Oftentimes, this means the government has to step in to provide such goods.
Of course, many types of information are in fact excludable. Keeping information secret or keeping access limited to a secure digital or physical environment is one solution to create excludability. While this may work for the Coca-Cola recipe and for data sets and algorithms, it does not work for information which must necessarily be exposed to create value, as is the case for works of art and literature and for news. From an economic perspective, copyright protection of literary and artistic works can be seen as a legislative intervention to make such works excludable, both in their physical form and in their digital form, and by doing so to create a potential market with room for private provision and for recouping investments.
Copyright protection applies to the free, creative choice of words – the journalist’s individual expression – forming a news article printed in a newspaper or online.Footnote 9 Recordings of radio and television broadcasts are also protected as intellectual property. The information itself, however – the news facts that are disclosed in an article or a broadcast – are unprotected.Footnote 10 Therefore, news facts possess two defining characteristics of public goods: non-rivalry and non-excludability.Footnote 11 With that public good character comes the perennial private investment puzzle for news.
18.2.2 Externalities of News
What makes the public good nature of news facts more pressing is that news is important in a democratic society. News consumption by an individual has benefits that stretch beyond that person. In economic terms, this is referred to as positive externalities of news consumption.Footnote 12 Chandra and Kaiser write: ‘Policymakers have long recognized that newspapers have a unique role in the civic discourse of a country, and have important consequences for informing the citizenry, encouraging electoral participation and providing a check on powerful forces in government and business.’Footnote 13
Arguably, there are positive externalities of news production as well. This argument relates to the aforementioned public good nature of news facts and the lack of intellectual property protection for facts as such. However, it focuses on the news production sector. While uncovering news facts may require substantial and long-term investments in research and sources, other news media are free to report on these facts themselves once they have been made public and can engage in ‘cream skimming’ by making selective investments in follow-up news production that repeats the facts and avoids copying of the individual expression used in the original news story.Footnote 14 In view of the positive externalities of news consumption, this is a good thing – the more media can report about certain news facts, the more people can learn about them – but it undermines the incentives to invest in news production, in particular when it involves costly investments such as foreign correspondents and investigative journalism.
Strömberg discusses the ‘mounting empirical evidence that the media play a key role in enhancing political accountability.’Footnote 15 One such piece of evidence is a study by Oberholtzer-Gee and Waldfogel, who find that Hispanic voter turnout in the United States increased by 5–10 percentage points in markets where Spanish-language local television news became available.Footnote 16 Strömberg concludes that ‘[a] number of surveyed studies find that an increase in media activity is associated with better policy outcomes, some of which use methods that reasonably convincingly identify causal media effects. There is some evidence that these media effects occur because the media transmit information to voters, which improves both the incentives and the selection of politicians.’ Next to such positive externalities of news consumption through better-informed voting behaviour, one may add the positive effect of informing others at the water cooler.Footnote 17 At a macro level, Parcu and colleagues cite graphs that show the striking correlation between the economic freedom of countries and their freedom of the press, and between the corruption index of countries and their freedom of the press.Footnote 18 More press freedom goes along with more economic freedom and less corruption. While proving causality in such analyses is difficult and effects may well run both ways, it remains relevant to acknowledge such correlations.
Going one step further, Ali argues that journalism, in particular local journalism, is not just a public good but also a merit good, which individuals should have access to regardless of their preferences and consumer behaviour.Footnote 19 His arguments for taking this more normative stance mostly coincide with the points made here about the positive externalities of news consumption.
18.2.3 Commercial Financing of News
As mentioned in Section 18.2.1, the public good nature of news facts, in combination with the lack of intellectual property rights over them, in itself tends to lead to underinvestment in news production. On top of that, positive consumption externalities tend to lead to underinvestment in the consumption of news, as consumers take these externalities into account insufficiently in their own decision making. In economic terms, this leads to a willingness to pay for news, both in money and in time, that is lower than is desirable from a welfare perspective. The demand curve for news, which news producers and news publishers have to engage with to recover their investments in news production, insufficiently reflects consumption externalities.
These observations are, however, not new. In fact, news media have been confronted with this private investment puzzle since time immemorial. Besides public funding or subsidisation, which will be dealt with in Section 18.2.4, there have been three main market-based mechanisms to address this issue.
The first mechanism is cross-subsidisation from advertising. The concept is simple: advertisers want their ads to attract attention and are willing to pay for access to places that have consumer attention. News media are one such place where consumers spend time on a daily basis. This enables news media to charge advertisers for access to attention and to use the receipts as a supplementary (or their only) source to cover their investments in creating news. As it is sometimes phrased by cynics: ‘The news is printed on the backside of the ads’.
Although advertising has been a staple ingredient of news media for ages, the underlying concept of a two-sided market, where news media serve as a platform enabling interaction between two groups of interest, consumers and advertisers, has only been theorised in the economic literature since around the start of this century.Footnote 20 A more recent overview applied to media markets is provided by Anderson and Jullien.Footnote 21 One factor that determines the outcomes in different settings is whether or not consumers multi-home: that is, whether they are loyal to one platform/news medium or combine many in their media diet. Another is the extent to which consumers dislike (derive disutility from) advertising. In particular in audio and audiovisual media, ads that cannot be avoided or skipped are likely to be perceived negatively by consumers, and the more consumers dislike advertising, the more the advertising side of the market will have to cross-subsidise the news consumers. In printed media, research is not conclusive about whether and to what extent consumers dislike ads.Footnote 22
The second is the friction of time. This is reflected in Shapiro and Varian’s bon mot: ‘Information is like an oyster: it usually has the greatest value when it is fresh.’Footnote 23 While there is no copyright protection of news facts, there is value in being the first to bring them. For as long as other news media have not started reporting about the same news facts, this grants the initial news medium short-lived monopoly power over the news item, which can to some extent be monetised by selling newspaper copies or access, or by attracting more viewers that allow for generating more ad revenues. Although this does not resolve the underinvestment caused by positive externalities of news consumption, it does to some extent – and for a short period – mitigate the public good nature of news facts.Footnote 24
The flipside of this mechanism is that reducing this friction will also change the proposition of investing in news production. This was the subject of the famous historical case International News Services c. Associated Press from 1918, in which Associated Press (AP) complained that International News Services (INS) was pirating its news items. INS claimed it purchased AP’s newspapers legitimately on the east coast of the USA. Exclusive news information in these papers was wired to the west coast so that INS could bring the news there simultaneously with AP. The Supreme Court of the United States ruled that this was a case of unfair competition and that AP’s news could only be used as ‘tips to be investigated’.Footnote 25 Although this case is no longer a binding precedent, it illustrates how a reduction of time friction is relevant to the business of news producers. By only allowing AP’s news to be used as leads for further investigation, the Supreme Court basically reintroduced the friction that the telegraph had removed.
The third mechanism used by traditional media is bundling.Footnote 26 Economic theory offers various motives for providers to bundle. For example, offering products that are often purchased together in packages can reduce transaction costs: payment and selection only need to be made once. It can also be attractive to bundle products that are complementary in terms of their utility value and where demand for one increases demand for the other, such as left and right shoes, or sushi and soy sauce. A third, intuitively less obvious reason for bundling is described in detail by Adams and Yellen:Footnote 27 bundling can be a very effective way of skimming consumer surplus when the preferences of individual consumers for the components of a package are negatively correlated or unknown. In many cases, bundling is then considerably more profitable than charging prices that maximise profits per component. Thus bundling general interest news, sports news, financial news, and fashion news in one print newspaper or one subscription can generate more revenues than selling access to these areas of interest separately.
18.2.4 Public Financing of News
Public funding or subsidisation was already briefly mentioned in the previous section. It follows from the public good character of news facts and the positive externalities of news consumption that some level of public financing or subsidisation of news production and news consumption will be welfare-enhancing. Also, one can think of public provision of news media as analogous to the public provision of other public goods.
Indeed, both direct and indirect subsidies for, as well as public provision of, news media are common practice in most EU Member StatesFootnote 28 and also outside Europe.Footnote 29 Parcu and colleagues discuss direct and indirect subsidies as means of state support for media companies in the European Union, as well as public provision of media.Footnote 30 As a form of indirect subsidy, they mention the reduced VAT rate that applies to (news) media in many countries. State advertising can also function as a form of state support for media.Footnote 31
Naturally, the hard question is what level of financing, subsidisation, or public provision is optimal, and how to implement this without weakening the incentives for private provision and recoupment of investments. Parcu and colleagues discuss various drawbacks and risks of public support mechanisms and public provision of news media. Importantly, selective subsidies or support for certain news media and not for others can distort competition in the media landscape. The same applies to public provision of news media in a market where there is also privately funded supply. This can make the business case for independent and unsubsidised news media even more problematic. Also, support can be used to gain undue editorial influence and thereby undermine the independence of news media. Public provision of news can raise similar issues concerning independence, or may even be perceived as an intervention leading to news that lacks independence.Footnote 32 One can add to these concerns the possible risk that public service television and radio funding may be poorly targeted, as the money can also flow to entertainment productions with no wider public value.
18.3 Disruptions by Aggregator Platforms and Generative AI
This section analyses how the rise of aggregator platforms and generative AI affect the analysis in Section 18.2 and in particular the three market-based mechanisms – cross-subsidisation by ads, time frictions, and bundling – that traditional news publishers have used over time to overcome the public good problem of news.
18.3.1 Disruptions by Aggregator Platforms
Aggregator platforms compete with news media for the attention of consumers with an interest in news. To the extent that consumers have a time budget for news that is (more or less) fixed, or insofar as consumers want to inform themselves about specific events, this implies that aggregator platforms will cause a substitution of consumer attention away from news media. Less time spent on traditional news media will lead to less advertising revenues for them, even if the aggregator platforms themselves do not run ads. If they do run ads, the effect on advertising revenues for traditional news media is worsened.
A potential opposite mechanism assumes that consumers’ time budget for news is not fixed and that aggregator platforms generate additional traffic to traditional news media websites, which they can monetise in the form of ad revenues. Note that this latter effect would likely enhance positive externalities of news consumption. By extension, aggregator platforms may direct consumers to news media that they are not aware of or would not go to spontaneously. By doing so, they can lead to more variation in consumers’ media diet, thus enhancing media consumption plurality.Footnote 33
In an experimental study, Dellarocas and colleagues primarily find evidence for the first, negative, effect on news publishers. They write: ‘the more information is provided by the aggregator, in the form of longer snippets and accompanying images, the more time users spend on the aggregator and the less likely they are to click through to original articles. One way of interpreting our findings … is that anything that aggregators display in addition to an article’s headline, decreases click-through rates.’Footnote 34 Furthermore, the study finds that snippets placed high in a topic group are more likely to be clicked. This gives aggregators ranking power over traditional news media.
In addition to enhancing competition for consumer attention and advertising budgets, aggregator platforms also unbundle traditional news media by selectively choosing and presenting ‘snippets’, typically consisting of a headline, a short excerpt and a link, based on their own algorithms and often making use of user data. This implies that the traditional strategy of bundling different types of news to maximise revenues is circumvented by aggregator platforms as well. Lastly, the time friction that traditional news media benefit from is also removed, in the sense that an aggregator platform can publish such snippets instantly once an article is published.
In sum, aggregator platforms undermine the three market-based mechanisms mentioned in Section 18.2.3 that news publishers have used over time to overcome the public good problem of news and recoup private investments in news production. At the same time, these aggregators do not make investments in news production themselves.
By doing so, aggregator platforms aggravate the downward trend in advertising revenues that has long plagued news publishers. Their ad revenues have declined dramatically in the past few decades in most developed countries, especially those of local news outlets,Footnote 35 in sync with the rise of the Internet and large online platforms. Note that a decrease in newspaper advertising revenues due to the emergence of an alternative technology is not uncommon. Chandra and Kaiser mention that in the United States, newspaper advertising revenues declined by 28 per cent between 1929 and 1941, linked to the mass introduction of the radio in the 1930s, while the introduction of television in the 1950s led to a further decline.Footnote 36
However, the current decline is different in two ways. First, it is a result of cannibalisation of ad revenues by media that themselves invest little or nothing in news production. Second, the decline appears to be of unprecedented proportions. In the United States, newspaper advertising revenues are reported to have declined by more than 80 per cent between 2000 and 2020: a drop from 48.7 to 8.8 billion USD.Footnote 37 In the EU (plus the UK), advertising revenues in newspapers and magazines fell by almost 60 per cent between 2010 and 2020.Footnote 38
A simultaneous – albeit slightly less dramatic – drop in paid subscriptions leads to a double dip in revenues. In the United States, daily newspaper circulation (print plus digital) in 2022 was about a third of its fairly stable level between 1960 and 1990.Footnote 39 In Europe, daily circulation declined by 23 per cent between 2015 and 2021. A steady growth in digital circulation could not make up for a sharper decline in print circulation.Footnote 40
18.3.2 Disruptions by Generative AI
A second and even more recent disruptive force comes from generative AI. As was mentioned in the introduction to this chapter, the role of generative AI is twofold. On the one hand is AI journalism: the use of generative AI tools by traditional news media to produce news articles, reducing the involvement of human journalists in the process of news production. In March 2025, the Italian newspaper Il Foglio took this to the extreme by publishing the first fully AI-generated (four-page) edition.Footnote 41 While one may have serious concerns about the effect this will have on employment for journalism and the very nature of journalism (see Chapter 21 by Martin Senftleben in this volume), from an economic perspective the use of AI by traditional news media is not a disruptive development in terms of market failure. As long as news media continue to make editorial choices, do fact-checks, and take editorial responsibility for their content, as the editor of Il Foglio did in their experiment – and which in the foreseeable future will require at least some involvement of human journalists – it can be seen as an innovation or tool that reduces the costs of news production at a time when efficiency gains are direly needed.
The second role of generative AI concerns its use by third parties to create news articles based on other news articles and information the AI system has access to: AI news digests. There can be various ways in which this may enhance or improve news consumption and by doing so contribute to the positive externalities of consumption. Think of a generative AI tool that one can ask to write a 2,000-words summary of today’s news, based on the websites of the Washington Post, The Times, the Frankfurter Allgemeine Zeitung, and Le Monde, and to read it out loud on one’s commute – a tool that one can subsequently ask to elaborate one of the items or to rephrase it in simpler language. Clearly, this can increase news consumption and comprehension, help reach new markets of news consumers, and tailor news in a way mass news media cannot. Different from the use of generative AI by news media to reduce production costs, this can alter the way in which news is consumed much more fundamentally and disruptively than aggregator platforms did.
There is no need to stress that this is hardly science fiction. In January 2025, Apple temporarily suspended an AI tool that provides short summaries of news stories because it was found to be inaccurate – hallucinating – in some of its reporting,Footnote 42 but it is likely to be only a matter of time before such start-up problems will be resolved. Online, there are already other tools on offer that claim to provide similar services. Bearing in mind that human journalists also make mistakes, there will be a margin of error for such tools that is acceptable for practical purposes.
These developments are likely to be even more disruptive than aggregator platforms for investment incentives in news production and by extension for traditional news media. Generative AI tools can harvest news facts that are not protected as intellectual property without any interest in the text of the article that is protected. Unlike aggregator platforms, AI news digests have no need to apply restrictions to the length of snippets to avoid violations of copyright. Instead, such tools can generate independently readable articles of any desired length. As long as they explain news facts in their own words – and avoid substantial copying of protected individual expression of source articles – no copyright infringement occurs. Therefore, the public good market failure will be more pressing than ever.
Some mitigation could be provided by access fees that traditional news media may charge to AI tools, but it is unlikely that such fees will be of any significance towards recouping the investment required for the production of news. Current news is often available via different sources, many of which do not have paywalls or other access restrictions, as a result of which news media will have limited negotiating power individually. In addition, and bearing in mind the positive externalities of news consumption, it would be socially undesirable if all news media were to place all their content behind paywalls.
To make things worse, the market-based financing strategies traditionally employed by news media will be further eroded by AI news digests. They will compete with news media for the attention of consumers and for advertising revenues even more fiercely than aggregator platforms. Meanwhile, time frictions that used to provide some short-lived monopoly power over news items will be negligible and bundling will be irrelevant. Combined, these effects will further undermine the investment incentive for news production.
18.4 Potential Remedies
As was noted in Section 18.2.3, the problematic business case for investment in news production, made even more problematic by the positive externalities of news consumption, is in itself not new. Public support for news production and public provision of news media were discussed in Section 18.2.4 as possible solutions. These have been around for centuries, despite their possible drawbacks in terms of market distortion and the risk of compromising the independence of the media. News aggregator platforms and emerging AI news digest tools are rapidly making this case for public support more pressing as they further erode the investment incentives in news production. Several authors have stressed that in particular local journalism is vulnerable to these disruptive forces.Footnote 43
Against this background, it is advisable to create public funds that different media can benefit from, or, where such funds already exist, to increase them.Footnote 44 Such funds should be managed by boards that are independent from political power to ensure preserving the independence of news media, should be platform-neutral and should be open to all parties that meet certain general eligibility criteria, in order to avoid market distortion.
Alternatively, subsidies could be directed toward consumers in the form of an individual news budget. While such a solution has the potential advantage of being demand-driven instead of supply-driven, it will have to answer the same difficult question of media are eligible for spending one’s news budget on, and which are not. A drawback is that such a solution is likely much more expensive to implement practically. As another alternative to subsidising news creation or consumption directly through funds, one could consider tax breaks or exemption from VAT in the consumer market. Again, the eligibility question of which media could benefit from this would need to be addressed.
Of course, funds or tax breaks will have to be financed by other sorts of taxes. It may seem natural to consider introducing levies or taxes on the use of aggregator platforms and generative AI tools for this purpose. However, some caution is warranted: in addition to raising money, such levies or taxes would make the use of these platforms and tools less attractive. To the extent that this would reduce news consumption, it would thereby also reduce its positive externalities.
Open access