The supply side of media bias

In my ECONS102 class, we cover economic explanations for media bias. Drawing on past research from Matthew Gentzkow and others, we demonstrate that there are demand-side explanations (media bias arises because of a bias in the preferences of the news-consuming public) and supply-side explanations (media bias arises because media firms segment the market, and focus on different segments). I even shared my teaching approach in this article published in the Journal of Economic Education in 2023 (ungated earlier version here).

I'm always on the lookout for more research in this space, which can add to our understanding (and potentially flesh out more details for my teaching of media bias). One example is this 2024 working paper by Tin Cheuk Leung and Koleman Strumpf (both Wake Forest University). They focus on the supply side of the media market, and their method:

...evaluates the factors which shape article duration on a newspaper’s digital homepage. Homepage articles have a privileged position, as it is the starting point for most readers which in turn drives views and shares of the featured articles. One of the most important editorial decisions is selecting how long each article remains at this prized location. After adjusting for demand-side effects using reader engagement data, we use prolonged homepage presence of ideologically slanted articles as indicative of supply-side biases.

Leung and Strumpf focus on articles in the New York Times (a more liberal media outlet) and the Wall Street Journal (a more conservative media outlet), using data from August 2021 to May 2023 for the NYT, and from October 2022 to May 2023 for the WSJ. They also look at data from Twitter (about 22 million tweets for the NYT sample, and about 2 million tweets for the WSJ sample). First, they use textual analysis to determine the sentiment and political leaning of each article from each of the media outlets. From this, they find:

...a liberal bias in NYT articles, with an average pro-Democrat score of 0.6 (where 1 represents full pro-Democrat alignment and 0 indicates complete pro-Republican alignment), while WSJ articles exhibited more neutrality, with an average score of 0.5.

Next, Leung and Strumpf look an article's presence on the homepage of each media outlet, and how that relates to tweet count (as a proxy measure of the article's popularity). Specifically, they evaluate:

...the causal impact of homepage presence on tweet count, using time-of-day homepage updating patterns as an instrumental variable... There are distinctive hours of the day in which there are large purges of homepage content, likely associated with editor shift changes. These hours occur throughout the day and so are unrelated to article newsworthiness though they do shape which articles are on the homepage. After instrumenting we find that tweet counts for articles increased significantly when featured on the homepage (35% increase for NYT articles and 162% for WSJ articles).

Unsurprisingly then, articles that appear on the homepage are more popular, receiving far more attention from the public. Next, Leung and Strumpf apply survival models to investigate the factors associated with how long an article remains on the media outlet's homepage. First, unsurprisingly, popularity matters:

A key finding from our analysis is the significant impact of demand-side factors. There is a strong negative relationship between tweet counts at t−1 (a proxy for article popularity) and the hazard rate (the likelihood of an article being removed from the homepage). For the NYT, a one standard deviation increase in log tweets (1.37) leads to 2.5% decrease in the hazard rate. This effect is even more pronounced for WSJ, where a one standard deviation in log tweets (1.09) leads to a 10.0% decrease in the hazard rate. This indicates that more popular articles, as measured by tweet counts, tend to remain on the homepage longer, underscoring the role of reader demand in determining homepage duration.

The 'hazard rate' is the rate at which an article is removed from the homepage. So, a lower hazard rate means that the article remains on the homepage for longer. The demand side (popularity) matters, but the supply side matters as well:

Supply-side factors are particularly influential for the NYT. A one-standard-deviation increase in an article’s pro-Democrat score, indicating its political slant, is associated with a 2% decrease in the hazard rate, which translates into approximately an additional 30 minutes of homepage exposure. This relationship holds across different news tones, suggesting that a wide range of NYT’s editorial decisions on homepage duration are influenced by the article’s political alignment....

For the WSJ, the influence of supply-side factors is generally negligible, both overall and for each news tone. The pro-Democrat scores of articles do not significantly affect their duration on the homepage, contrasting with the NYT, where political slant plays a noticeable role.

So, the NYT's approach to its homepage is affected by the supply side, but the WSJ's approach is not. Leung and Strumpf suggest that:

This suggests that the WSJ might prioritize profit-maximizing objectives over ideological considerations (recall demand-side effects are also larger for the WSJ).

That would be consistent with the models I teach in my ECONS102 class. A profit-maximising media firm should base its' decision about where to locate its content on the liberal-conservative scale based on the relative demand from the news-consuming public, and not at all based on the editorial preferences of the owners or editors. Leung and Strumpf are essentially suggesting that this is true of the WSJ, but not the NYT.

Interestingly though, sentiment matters as well:

The sentiment of an article, captured through abstract sentiment scores, is notably linked to its duration on the homepage. For the NYT, a one standard deviation increase in the abstract sentiment scores is associated with a 2% decrease in the hazard rate, indicating that more positively toned articles are likely to remain on the homepage longer. The effect is even more pronounced for the WSJ, where a similar increase in sentiment scores corresponds to a 4% decrease in the hazard rate.

Whatever happened to 'if it bleeds, it leads'? Since more positively framed news stays on the homepage for longer, that old mantra may be on the way out. Overall, it seems that both demand and supply sides affect which articles receive prominence on the homepage of the NYT, and that for the WSJ it is mostly the demand side. In the latest version of their paper [*], Leung and Strumpf conclude by noting the importance of understanding the supply side of media bias, in an era when local newspapers are closing, increasing the risk of media bias.

[HT: Marginal Revolution, last year]

*****

[*] In the January 2024 version of the paper, which is the one I read some time back, Leung and Strumpf had clearly used ChatGPT (or some other LLM) to write the conclusion, with hilarious results, including a 'rich tapestry', a 'symbiotic relationship', an 'intricate balance', a 'multifaceted realm', 'delving', and a 'deep dive'. Thankfully, the latest version appears to have resolved this issue.

The impact of Fox News on American democracy

In yesterday's post, I noted a number of opportunities for research on the economics of social media. At least one of those opportunities intersected with the impact of traditional media. So, I was interested to read this new article by Elliott Ash, Sergio Galletta, Matteo Pinna (all ETH Zurich), and Christopher Warshaw (George Washington University), published in the Journal of Public Economics (open access). They look at the impact of Fox News Channel on political ideology of voters, and voting outcomes in the US. Given the well-known right-wing nature of Fox News, it is reasonable to wonder whether it is having a political impact.

Identifying the causal impact of a television channel is somewhat challenging, because where viewership is higher that might be because there are more right-leaning voters. So, the causality might run from voters to viewership, rather than the other way around (reverse causality). However, Ash et al. make use of the fact that the channel number assigned to Fox News varies across markets, and that assignment is random (or, at least, it isn't related to the partisanship of the population in a particular television market). So, Ash et al. use channel position as an instrument for viewership (I'll come back to this point later). They then look at the political preferences of voters using data from:

...the 2000 and 2004 National Annenberg Election Survey (NAES) and the 2006–2020 Cooperative Congressional Election Study (CCES) surveys. Overall, we have data on the preferences of approximately 661,000 Americans.

They also look at the effect on presidential and down-ballot (senate, gubernatorial, and house) elections using a variety of election data sources. They find that:

...from 2006–2008 onwards, a lower Fox News Channel (FNC) position correlates with an increase in self-identified Republican viewers, significant at the 10% level initially and 5% in recent years. A one-standard-deviation drop in FNC’s channel position corresponds to a roughly one-percentage point rise in Republican self-identification... A one-standard-deviation decrease in FNC’s channel leads the average American’s ideological position to shift .03-.04 standard deviations to the right in recent years...

Looking across time, the results are not statistically significant (at the 5 percent level) until 2009-2012, or later, depending on the measure. Turning to presidential elections, Ash et al. find that:

Initially, in the 2000 and 2004 elections, FNC’s impact was minimal, likely due to its growing viewership. By 2008, a one-standard-deviation decrease in FNC’s channel position correlated with a 0.32 increase in the Republican vote share...

And on down-ballot elections:

Overall, the effects of FNC in down-ballot elections are qualitatively similar to, though less precise than, those in presidential elections. In House elections, we see a positive (Pro-Republican) coefficient on FNC for Republicans’ two-party vote share in 2012, and it becomes statistically significant starting 2018. Since around 2012, counties with one-standard-deviation lower FNC’s channel position have about .6 to 1.6 percentage point greater shares in Republican vote.

In Senate elections, we find a positive coefficient starting in 2006, which becomes statistically significant starting in 2014.23 Since then, there have been relatively consistent year-to-year effects of around .6 to .73 percentage points. In other words, a one standard deviation shift in FNC’s channel position increases Republican Senate candidates’ vote share by over half a percentage point.

In gubernatorial races, FNC had a small and statistically insignificant effect until the latter half of the 2010s. In recent elections, however, the effect of FNC in gubernatorial races is similar to Senate races — with a Republican vote share about .5 percentage points higher...

So, it is clear from the results that Fox News Channel is driving a rightward shift in the voting public, and this is having a clear effect on both presidential and down-ballot elections. Ash et al. conclude that:

Given the estimated effect sizes on presidential elections, for example, Fox News could have easily tipped the scales for Donald Trump in 2016.

Perhaps Donald Trump should think himself lucky that his 2022 feud with Fox News didn't escalate too much?

However, there is some reason for scepticism about these results. Although Ash et al. make a good case for their use of channel position as an instrument for viewership, it turns out that they didn't actually run a full instrumental variables analysis:

Ideally, we would provide first-stage and two-stage least-squares (2SLS) results for all years in our analysis. However, there are limitations in estimating and interpreting the 2SLS results. First, we only have data on both the endogenous regressor (FNC ratings) and channel positions for 2005, 2006, 2008, and 2020. Second, the first stage provides evidence that the typical variation induced by the channel positioning is somewhat limited, suggesting that the size of the 2SLS coefficients would be, by construction, unreasonable.

...our main analysis focuses on the reduced form, where the outcomes (e.g., vote shares) are regressed directly on the instrument (FNC channel position)...

So, while they have a good instrument, a lack of data prevents them from making the most of it. And so, I think we would need further evidence before we can conclude that these results demonstrate a causal effect of Fox News Channel on political outcomes in the US. It seems to me that the main holdup to doing a full instrumental variables analysis is not having all the viewership data from Nielsen. So perhaps some wealthy research institution needs to buy that data?

The economics of social media and opportunities for future research

There has been an explosion in research into social media over the last decade, as you might expect. In a new article published in the Journal of Economic Literature (ungated earlier version here), Guy Aridor (Northwestern University), Rafael Jiménez-Durán (Bocconi University), Ro’ee Levy (Tel Aviv University), and Lena Song (University of Illinois Urbana–Champaign) document the trend. Here's Panel A of Figure 1 from the paper:

The increase in published research is clear (the "General interest publications" they refer to are more-or-less the top general journals in economics). Aridor et al.'s article reviews this large and growing literature, highlighting what is already known about the economics of social media, as well as where they see gaps suitable for future research. The focus of the review is often narrowed to a case study related to social media and politics, which while important in itself, is not the only important factor associated with social media. So, there may be additional research opportunities outside of those that are identified in the review.

Aridor et al. define social media as "two-sided platforms that primarily host user-generated content distributed via algorithms, while allowing for interactions among users". They then structure their review to follow the "flow of content" in social media, starting with content production (generated by users), then content distribution (including both 'organic content' and advertising, distributed by platforms using algorithms), and finally content consumption (again, by users).

The review is difficult to summarise, so if you are interested, I encourage you to read it in detail. In this post, I just want to highlight some of the research opportunities in this space (but not all of the opportunities that Aridor et al. describe - I'll focus on what I think are the more important or interesting ones).

First, in terms of content production, Aridor et al. note that:

...content creation is increasingly viewed as a viable career or income source... A natural next step to the existing evidence on the elasticity of the content supply curve concerns the labor economics of this activity, studying questions such as the effects of unions for content creators or whether monetary incentives crowd out nonmonetary motives. Beyond ad revenue–sharing programs, other monetary incentives that have been increasingly used by platforms (for example, allowing users to subscribe to producers) remain understudied, perhaps due to missing data.

Also:

More research is needed on how social media algorithms could affect the production of other types of content beyond news. For example, it has been argued anecdotally that TikTok is driving songwriters to focus on brief, danceable 15-second snippets.

In general, TikTok has probably been under-studied relative to other platforms, particularly given its large user base. Turning to misinformation, Aridor et al. note that:

The literature has mostly focused on the sharing, as opposed to the production, of misinformation. One potential reason for this imbalance is the role of resharing in diffusing misinformation... An important gap in this literature is to understand the determinants of the production of misinformation, beyond the sharing of existing articles.

And on fact-checking (which may now be more difficult to research given that Meta is giving up on fact checking):

One gap in this literature is to disentangle a potential dual role of fact-checking interventions, which affect not only the users’ perceived veracity of the content they are about to share but also the perceived likelihood that they will be fact-checked by the platform in the future.

Aridor et al. then turn to sanctions on users, and note that:

Besides crowd-sourcing, platforms conduct other content moderation measures at scale, such as down-ranking or removing posts, banning groups, and suspending user accounts... However, more research is needed to understand the causal effects of these “harder” interventions on the production of misinformation and the mechanisms through which they operate, whether they crowd out fact-checking efforts by the users, and the net welfare effect of sanctions.

And on counter-speech to reduce the toxicity on platforms:

An open question is what determines the equilibrium provision of counterspeech and how to incentivize users to provide this public good (similarly to fact-checking).

Finally on content production, Aridor et al. note that:

More evidence is needed on the connection between content moderation and advertising. Specifically, there is limited research examining how content moderation influences advertisers and, conversely, how advertising dynamics influence content moderation decisions... more evidence is needed to understand the effect of hate speech and other types of content on user interactions with advertisements and whether content moderation policies can alleviate any potential negative effects.

There seem to be a number of promising research opportunities in the content distribution space (and that likely primarily relates to a lack of access to the algorithms, which is why these questions haven't already been answered). Specifically, Aridor et al. note that:

Future research could examine what drives demand for social media content and when and to what extent content consumption is driven by algorithms. 

And importantly:

An open question is how social media algorithms can optimally increase social welfare and what government incentives can encourage them to do so.

That last question is a big one, with huge potential for policy impact. I would be very surprised if there weren't already concerted research efforts in that space. Also:

...more research is needed on how users decide which pages to follow (for example, the accounts of media outlets or politicians), since those pages may be driving segregation.

Then, turning attention back to advertising, Aridor et al. note that:

Economic analysis of advertising... posits that advertising primarily works through the following channels: shifting beliefs through information (for example, product awareness, attribute information) or directly shifting consumer preferences (for example, increasing affinity to the brand)...

Empirical research suggests several unique aspects of social media advertising where these mechanisms interact with the “social” aspect of social media.... more work is needed to understand the relative role of each of the different mechanisms for ad effectiveness and how they interact with both the increased targeting abilities and the social aspect of social media advertising.

 And finally:

...such a large amount of money is spent on digital advertising in national election campaigns is puzzling and deserves additional research, since either these ads are more effective than current research indicates or researchers are wrongly inferring the objectives that campaigns pursue with these ads (for example, fundraising rather than voter persuasion).

That question relates to some of my own earlier (and most cited) research on the impact (or rather, non-impact) of the number of social media followers on politicians' chances of being elected (see here, or here for an earlier ungated version). Aridor et al. also note that:

One direction for future work, even for studying these issues within the United States, is to better understand the impact of these types of ads in local elections, which is precisely where we may expect that social media advertising could have a larger impact.

I can say that in some follow-up work on local elections in New Zealand, which we haven't published, there was no significant effect of social media (which didn't surprise us at all, given the earlier results in national elections I pointed to above).

Aridor et al. then discuss opportunities in researching content consumption, and note that:

There are several avenues for future research. First, given the complex and fast-evolving nature of social media consumption, descriptive evidence detailing consumption behavior would be valuable... Second, while existing research highlights self-control problems among American adults, it is policy relevant to quantify the extent of these problems in the younger population. Finally, future work could look within platforms and quantify how different design features influence what and how users consume. Certain features (for example, content format or algorithms) may exacerbate self-control problems... Defining the key product characteristics and quantifying their effect on consumer choice is an important step forward in understanding the welfare implications of consumption.

And importantly:

There is convincing evidence that social media use—particularly the exposure to toxic content—can lead to offline hate crimes...

There is also some evidence that government regulation akin to a Pigouvian tax can mitigate this externality... Additional research is needed to study whether these policies have unintended consequences such as the silencing of political dissidents. Lastly, future research could explore how content moderation affects other offline harmful actions besides violence (for example, self-harm)...

And finally:

There are several interesting directions for future work. The first is that given the large informational externalities from consumption... an unexplored question is not only to measure market power in terms of time spent, but also to think of media power as Prat (2018) does for traditional media. Of particular interest is understanding whether social media increases or decreases the media power of existing large media organizations. The second is to explore the implications of habit formation... for competition among social media platforms.

The overall takeaway from this review is that, while there is a large (and quickly growing) literature on the economics of social media, there is still substantial scope for future research that will have real policy and practical impact. I look forward to seeing many of these research questions addressed in the near future.

News coverage and mass shootings in the US

Does widespread media coverage of mass shootings increase the likelihood of future mass shootings? It seems intuitive that the answer could be 'yes'. Perhaps media coverage encourages 'copycat' behaviour. Perhaps, the prospect of fame motivates mass shooters, and regular media coverage provides the avenue to fame (or, rather, infamy). Maybe media coverage simply makes mass shootings seem 'more normal', as a response to some slight. Teasing out which of these explanations is correct is difficult, especially when you consider that the direction of causality isn't even established. Does media coverage lead to more mass shootings (as the theories outlined above suggest), or do more mass shootings simply lead to more media coverage?

Those are the questions that this recent article by Michael Jetter (University of Western Australia) and Jay Walker (Old Dominion University), published in the journal European Economic Review (sorry, I don't see an ungated version online), set out to address. Jetter and Walker obtain data on mass shootings (four or more victims) for the period from 2006 to 2017, from USA Today's Behind the Bloodshed database. To understand the extent of the problem, they highlight that:

...265 (or six percent) of the 4383 sample days from 2006 to 2017 experienced at least one mass shooting, whereas nine days saw two mass shootings.

Jetter and Walker then collect data on television news from the Vanderbilt Television News Archive. They count the number of news segments from ABC World News Tonight (the highest-ranked evening news programme in the US) that contain the terms 'shoot' (including 'shooter' and 'shooting'). They then filter out false positives manually (such as news items about the shooting of a movie). They find 490 news segments on shootings over the 2006 to 2017 period.

Their analysis relies on an instrumental variables (IV) approach, to avoid the reverse causality problem and obtain the causal effect of news on shootings. To achieve this, they need to identify an instrument that affects the amount of news on shootings, but doesn't directly affect the number of shootings. They suggest disasters in other countries, where large numbers of US expatriates live. Jetter and Walker collect data on disasters from the International Disaster Database, limiting the data to countries with at least 50,000 US emigrants. They note that:

In total, we capture 158 such disasters (132 earthquakes, 10 epidemics, and 16 events of volcanic activity) that span 950 days out of our 4383 sample days.

It turns out that disasters are a good predictor of shooting news (because more disaster news leaves less room in the news broadcast for shooting news). In their IV analysis, Jetter and Walker find that shooting news is:

...a positive, statistically significant, and quantitatively powerful predictor of mass shootings. The corresponding magnitude indicates a one standard deviation increase in shooting news (0.389) translates to an increase in the number of shootings by approximately 73% of a standard deviation.

Now, looking over the time after shooting news, they find that:

The derived coefficients decrease in magnitude and eventually turn statistically indistinguishable from zero after 3-4 weeks.

So, news coverage of shootings causes an increase in mass shootings, and the effect lasts for up to 3-4 weeks. The next question is, why? Jetter and Walker look further into the data, and find no causal relationship from news coverage to murder more generally. That rules out salience, or that shootings are seen as 'more normal' after news coverage of shootings. They find that the mass shootings are more likely to happen on or after the anniversary of other infamous mass shootings, which suggests a behavioural contagion (or 'copycat') effect. Finally, they find that mass shootings are no less likely to happen on days where there would be predictably less change of news coverage (such as the dates of the Super Bowl, Olympics, Academy Awards). That would seem to rule out fame as a significant motivator, since a mass shooter who wants to be famous would probably want to avoid those dates, and yet they don't. So, Jetter and Walker conclude that:

Taken together, the results... are consistent with a behavioral contagion model...

Mass shooters are copycats. And then, in terms of policy implications:

First, our results advise journalists to report less on mass shootings... Second, our results also explain (at least in part) why shootings sometimes cluster in short intervals... Thus, police and other security forces may be well advised to be alert on and after days of heightened media coverage of mass shootings.

In the Waikato Economics Discussion Group a couple of weeks ago, we discussed this paper. We decided you couldn't really replicate this analysis for New Zealand, as (thankfully) mass shootings are very rare. However, one student made mention of ram raids, which have attracted massive attention in the media, this year in particular (for example, see here). Is media coverage of ram raids driving up the number of ram raids in New Zealand? The arguments are very similar to those in the Jetter and Walker article. Whether ram raids are driven by media coverage is an interesting question, and one that may be well worth following up on in an Honours or Masters research project.

Does the media help the general public understand inflation?

The media has an important role in helping us to recognise and understand what is going on in the world around us. One of the main goals of my ECONS102 paper is to have students recognise how economic concepts and theories apply to things that are discussed in the media. So, when I read the abstract to this new article, by David-Jan Jansen (De Nederlandsche Bank) and Matthias Neuenkirch (University of Trier, Germany), published in the Oxford Bulletin of Economics and Statistics (I don't see an ungated version, but it might be open access in any case), I took note.

Jansen and Neuenkirch use data from around 1800 people in the Netherlands, over the period 2014 to 2017, and investigate whether their engagement with print media affected the accuracy of their perceptions of current and future inflation. They find:

...no real support for the idea that more-often informed members of the general public do better in understanding inflation. In fact, more frequent readership of some types of newspapers is associated with slightly less accurate inflation perceptions... Also, a set of cross-sectional regressions using data collected in 2017 finds no evidence that non-print media outlets, such as television or Internet, help in understanding inflation. Overall, our paper casts further doubt on the idea that media usage contributes much to knowledge on economic developments...

In other words, the authors argue that the media plays no informative role in helping the general public to understand inflation. However, I disagree for two reasons.

First, the paper is clear that their measure of media use is simply the proportion of newspapers (by type: 'quality' or 'popular') that each person in the sample reported that they used 'frequently' or 'very frequently'. This is hardly a measure of media engagement. It is a measure that mixes up the diversity of news sources the person engages with, with the intensity of that engagement. For instance, there is only one news source that I would report that I engage with 'frequently', so I would show up as a low media user in their sample (in fact, given that it is print media and I don't read print newspapers, I would actually show up as a non-user).

It isn't clear to me what a relationship between incorrect inflation perceptions and the proportion of print newspapers that people read frequently even tells us. It certainly doesn't tell us anything about whether the media helps the general public to understand inflation.

The second issue I have with the paper is their measurement of error in inflation perceptions (and in future inflation expectations). They look at the absolute value of the difference between actual and perceived inflation (or expected future inflation). Actual inflation was taken from the Dutch statistics agency, and perceived inflation was what the respondents thought inflation was in that year (and in the following year, for inflation expectations). The problem here is the use of the absolute value of the error. Let's say that paying attention to media sources makes the general public overestimate inflation. Some people would overestimate inflation, but would overestimate by more if they paid attention to the media. However, other people would underestimate inflation, but by less if they paid attention to the media. In the paper, it appears that most people overestimate inflation, but the conflation of these two groups still potentially creates a serious problem for the analysis.

This is a paper that looks at an interesting research question, but does so in such a way that it doesn't actually answer the research question. It would be reasonably straightforward to replicate this analysis in a more sensible way though, if the authors were willing to share their data (the Dutch DHS household data is freely available online, but the authors supplemented that with their own data collection).

And if the answer is that the print media doesn't help the public to understand economic developments, we could conclude that the public should spend more time reading economics blogs instead.

Natural disasters and news media bias

In ECON110, we discuss the economics of news media bias. In discussing bias, I explicitly discuss sensationalism (bias of the exceptional over the ordinary), recency bias (bias of the new over the status quo), and exaggerated influence of minority views (bias by “fair” representation of both sides of the argument). We then go on to discuss how the 'normal' operation of the media market introduces bias into the way news stories are reported. However, we don't really go into much detail on the consequences of new media bias.

A recent 2007 paper by Thomas Eisensee and David Strömberg (both University of Stockholm), published in the Quarterly Journal of Economics (ungated version here), provides an excellent example of the consequences of media coverage. In the paper, Eisensee and Strömberg look at how media coverage of natural disasters outside the U.S. affects the likelihood of disaster relief being given, using a dataset of 5,212 disasters over the period from 1968 to 2002. Specifically, they look at how news coverage of natural disasters is affected by coverage of other news stories (e.g. Olympics coverage), and how that affects whether or not the U.S. declares a natural disaster and provides aid. The news data is based on television evening news stories (ABC, CBS, NBC, and CNN).

They found that:

...2.4 extra minutes spent on the first three news segments (two standard deviations) decrease the probability that a disaster is covered in the news by four percentage points and the probability that the disaster receives relief by three percentage points. Recall that around 10 percent of all disasters are covered in the news and that 20 percent receive relief, and so the effects are sizeable... The estimated coefficients imply that a disaster occurring during the Olympics is 5 percent less likely to be in the news and 6 percent less likely to receive relief, on average.

The effects are especially large for disasters that are "marginally newsworthy" (those that would appear in the news if and only if there was little else happening that was newsworthy). More severe disasters (measured by the number of people killed or 'affected') are both more likely to receive news coverage and more likely to receive relief. However, the effects differ by the type and location of the disaster:

...we have computed the casualties ratio that would make media coverage equally likely, all else equal (controlling for the same factors as in the fixed effects regression). For example, for every person that dies in a volcano disaster, 38,920 people must die of food shortage to receive the same expected media coverage. The conclusion is that media induces extra relief to volcano and earthquake victims, at the expense of victims of epidemics, droughts, cold waves and food shortages...

The estimates suggest that it requires 45 times as many killed in an African disaster to achieve the same probability of media coverage as for a disaster in Europe. We conclude that media coverage induces extra U. S. relief to victims in Europe and on the American continent, at the expense of victims elsewhere.

In the Pacific, it requires 91 times as many killed to achieve the same probability of media coverage as for a disaster in Europe. In ECON110, we conclude that news media bias reflects the underlying bias in the news-consuming public (that prefers to consume news that is consistent with their own preferences and biases). I guess the U.S. news-consuming public cares less about people who die in famines or from natural disasters in the Pacific?

[HT: Marginal Revolution in July this year, but also see the excellent write-up in Our World in Data]

Rockonomics: The economics of popular music

I really enjoyed my time in Seattle, and especially the Museum of Pop Culture. The plane trip from Seattle to Portland (which we also visited before returning to New Zealand) seemed an opportune time then to read this 2006 chapter (with the same title as this post) from the Handbook of the Economics of Art and Culture (ungated earlier versions here and here), by Marie Connolly and Alan Krueger (both Princeton).

That chapter had been sitting in my must-read pile for about ten years (!), but for whatever reason consistently kept getting bumped a little lower down the pile. It is important to me because the economics of popular music provides a lot of good illustrations of the things we teach in first year microeconomics. So given that ECON100 and ECON110 both start B Semester lectures next week, in this post I'm going to take some brief quotes (taken from the NBER working paper version of the chapter) from the chapter to illustrate some of the things we will cover in those papers.

Besides that, the chapter includes a lot of interesting detail on the structure of the music economy. Consider these bits, which relate to the ECON110 topic on media economics:

...it is clear that concerts provide a larger source of income for performers than record sales or publishing royalties. Only four of the top 35 income-earners made more money from recordings than from live concerts, and much of the record revenue for these artists probably represented an advance on a new album, not on-going royalties from CD sales... 

If a band composed its own music, it will also contract with a publisher to copyright the music... The publisher usually takes half the royalties, and the composer receives the other half (some of which goes to the manager).

...bands receive relatively little of their income from recording companies. Indeed, only the very top bands are likely to receive any income other than the advance they receive from the company, because expenses – and there are many – are charged against the bands advance before royalties are paid out...

Record companies tend to sign long-term agreements with bands that specify an advance on royalties and a royalty rate. The typical new band has very little negotiating power with record labels, and the advance rarely covers the recording and promotion costs, which are usually charged to the band. Because fixed recording costs vary little with band quality, only the most popular artists earn substantial revenue from record sales...

[Quoting Jacob Slichter, the drummer for Semisonic]: If our CD was sold in stores for fifteen dollars, the band’s share of the revenue might be something between fifty cents and a dollar per CD.

We cover moral hazard in both ECON100 and ECON110. Moral hazard occurs when one of the parties to an agreement has an incentive, after the agreement is made, to act in a way that brings additional benefits to themselves at the expense of the other party. In relation to that Connolly and Krueger write:

Caves prosaically notes that, “From the artist’s viewpoint, a problem of moral hazard arises because the label keeps the books that determine the earnings remitted to the artist.”

So, the recording label engages in moral hazard because it provides additional benefits to the label, and because the artists cannot easily monitor what the label is doing when it estimates earnings and costs and what should be paid to the artist. There are also a number of points that Connolly and Krueger write in relation to pricing (which we cover in ECON100), including:

As an economic good, concerts are distinguished by five important characteristics: (1) although not as extreme as movies or records, from a production standpoint concerts have high fixed costs and low marginal costs; (2) concerts are an experience good, whose quality is only known after it is consumed; (3) the value of a concert ticket is zero after the concert is performed; (4) concert seats vary in quality; (5) bands sell complementary products, such as merchandise and records...

The price of a concert ticket is set lower than it would be in the absence of complementary goods, because a larger audience increases sales of complements and raises revenue.

Firms that sell complementary products need not necessarily profit maximise for any of those products individually, if they can profit maximise across the whole range of their products. And in terms of price elasticity (covered in ECON100):

...despite flat or declining tickets sales, total revenues (in 2003 dollars) trended upwards until 2000 because of price increases. Other things equal, these trends suggest the elasticity of demand was less than 1 before 2000. Since 2000, however, there has been a 10 percent drop in ticket revenue for these artists, suggesting that prices increases have been offset by a larger than proportional demand response.

When demand is relatively inelastic, a given percentage increase in price is associated with a smaller percentage decrease in quantity demanded, so total revenues (price x quantity) increases. This is what happened prior to 2000, but after 2000 demand appears to have been elastic, so that the percentage increase in price was more than offset by a larger percentage decrease in quantity demanded, meaning that total revenues (price x quantity) decreased.

Connolly and Krueger cover inequality as well (as we will in ECON110):

...concert revenues became markedly more skewed in the 1980s and 1990s. In 1982, the top 1% of artists took in 26% of concert revenue; in 2003 that figure was 56%. By contrast, the top 1% of income tax filers in the U.S. garnered “just” 14.6% of adjusted gross income in 1998 (see Piketty and Saez, 2003). The top 5% of revenue generators took in 62% of concert revenue in 1982 and 84% in 2003.

And on intellectual property rights (which we cover in ECON110):

How far does intellectual protection go? Are rights strong enough to encourage the optimal amount of innovation? The problem stems from the fact that musical compositions are nonrival goods, whose property rights, as laid out by Nordhaus (1969), generate a trade-off between under-provision of the nonrival good (with weak rights) on the one hand and monopoly distortions (when the property rights are strong) on the other. 

Nordhaus's characterisation of the trade-offs inherent in intellectual property is one of the key pillars of the ECON110 topic on intellectual property rights. There are other bits of interest, including ticket scalping (which we cover in both ECON100 and ECON110), signalling (also both ECON100 and ECON110), and superstar effects (which we discuss in ECON110). A few parts of the chapter are a little technical, but all of it is interesting, and there are lots of gems to take away. Some parts of the chapter are getting a little dated, but mostly it has aged well and if you like to see economics in action, I encourage you to read it.

The puzzle of newspaper pricing

In the simple economic model of supply and demand, when demand decreases the price decreases (except in some special cases such as when supply is perfectly elastic). Typically, the same holds true for other market structures such as monopoly - when demand decreases, the monopolist's profit-maximising price usually decreases.

That makes the recent experience of newspaper subscriptions increasing in price, in the wake of decreasing demand, somewhat of a puzzle. Fortunately, a 2016 paper by Adithya Pattabhiramaiah (Georgia Tech), S. Sriram (University of Michigan), and Shrihari Sridhar (Texas A&M) unpacks the puzzle for us, and ultimately it rests on the fact that this is a platform market (or a two-sided market) - a market where a firm brings together two sides (e.g. in this case the readers and the advertisers), both of whom benefit by the existence of the platform (the newspaper), and both of whom may (or may not) be charged (in this case, the readers are charged a subscription, and the advertisers are charged for advertising). As Nobel Prize winner Jean Tirole has noted, in platform markets it is common for one of the two sides to subsidise the other. To be more precise, the side of the market with relatively more inelastic demand (or greater willingness-to-pay for access to the platform) will subsidise the side of the market with relatively more elastic demand (or lower willingness-to-pay for access to the platform).

So, one potential explanation for the puzzle of increasing subscription prices for readers in the face of decreasing demand is that the decreasing number of readers reduces the value of advertising in the newspaper, which reduces advertisers' willingness-to-pay for advertising, which in turn reduces the optimal subsidy the newspaper will apply to subscriptions. And it is this explanation that Pattabhiramaiah et al. set out to test.

They use data for 2006-2011 from a top-50 regional newspaper in the U.S., supplemented by microdata from 5565 subscribers, to construct models of: (1) the subscribers' decisions about whether to subscribe (and which of three subscription choices to select); (2) the advertisers' decisions about whether to advertise (and whether to do so in classifieds, display, or inserts); and (3) the newspaper's decision about subscription prices. A bit of background first though:

During the period of our analysis (i.e., 2006-2011), 14% of households in the newspaper’s market subscribed to the focal newspaper...

Conditional on subscribing to the focal newspaper, 72.4% (71.6%) of readers within (outside) the core market opt for the Daily option. The corresponding numbers for the Weekend and Sunday only options are 5.4% (4.4%) and 22.2% (23.9%), respectively...

The Daily option witnessed the steepest price increase of nearly 77%, both within and outside the core market, while prices of the Weekend and Sunday only options also increased by 52% and 38%, respectively...

On average, across the three options, the newspaper’s circulation witnessed steep year-on-year declines within (outside) the core market of between 7-10% (2-6%)...

While display and inserts lost 57.7% and 43.4%, respectively during our analysis period, Classifieds ad revenues experienced the steepest decline of 88.3%...

Between 2006 and 2011, classifieds ad rates at the focal newspaper declined by 66%, possibly as a result of the growing popularity of Craigslist. The rates for display ads and inserts experienced smaller declines of 16.7% and 10.8%, respectively.

So, subscription prices for readers increased (and readership decreased), and advertising rates (the price advertisers pay) decreased and so did advertising revenue. However, that decrease in readership was likely to be both the result of decreased price (and so was the result of a movement along the demand curve), and the result of changing reader preferences to (a decrease in demand, or a shift of the demand curve to the left). This creates an identification problem (which I've written on before, here) - which of the movement along the demand curve or the shift in the demand curve has contributed the most to the change in price? [*]

Pattabhiramaiah et al. then used their model to:

...compare how optimal markups evolved between 2006 and 2011 in each of the three cases: actual markups (computed based on our model parameters), the case where we switch off the decline in readers’ preferences, and the case where we switch off the decline in the incentive to subsidize readers at the expense of advertisers.

They find that:

...within the core market, the decline in readers’ preferences accounted for between 8-21% of the increase in subscription prices. On the other hand, nearly 79-92% of the increase in subscription prices between 2006 and 2011 can be traced back to the decreasing incentive on the part of the newspaper to subsidize readers at the expense of advertisers.

So, their results support the argument that the puzzle is explained by decreasing subsidies from the advertiser side of the platform market to the reader side.

There are two final (statistical) points I want to make about this paper, which may leave us with some concern about the results. The first is illustrated by this quote from the paper:

We find that the correlation between the subscription of the local newspaper and the local subscription of national newspapers is 0.8, suggesting that it is not a weak instrument... Overall, these results suggest that the instruments, along with other exogenous variables, explain 83% of the variation in readership. Compared to the first stage regression with only the exogenous variables, but excluding the instruments, the proposed instrumental variables improve the R-squared by 12-14% for ad rates and 11-15% for readership. Therefore, we contend that we do not have a weak instruments problem.

What the hell? There are actual statistical tests that you can run for testing whether you have weak instruments (Stock and Yogo have a whole chapter on it here), but rather than report the results of those tests they obfuscate instead? On the basis of what they have written, we are left with little idea about whether their instruments do a good job or not (for more on instrumental variables, see my post here). The second issue is somewhat hidden in a footnote:

Our in-sample MAPD ranges between 17.4%-17.8%. The out of sample MAPD range between 12.1%-16.8%.

The MAPD (Mean Absolute Percentage Deviation) is a measure of the error in their model, and it is extremely unusual for a model to show a lower error on data that was held over for validating the model (out-of-sample data) than on data that was used to construct the model (in-sample data). After all, most models are constructed to minimise the in-sample error. So this should leave us a little concerned about their model (or their calculation of MAPD). Or perhaps it's just luck that their data does a better job of predicting the 2010-2011 period than the 2006-2010 period?

[HT: Marginal Revolution]

*****

[*] Pattabhiramaiah et al. use their data to first eliminate the alternative possibilities of increasing quality of the newspaper leading to an increase in price, or increasing marginal costs leading to an increase in price (in fact, they find that marginal costs actually declined over the period).

Book Review: Merchants of Doubt

I just finished reading the 2010 book by Naomi Oreskes and Erik M. Conway, Merchants of Doubt. The subtitle is "How a handful of scientists obscured the truth on issues from tobacco smoke to global warming" and is a pretty fair summary of the book. The authors have completed what appears to be a very thorough investigation of several scientists whose names keep appearing as combative opponents against the broad consensus views of scientists. The authors summarise their findings on pages 6-7:

In case after case, Fred Singer, Fred Seitz, and a handful of other scientists joined forces with think tanks and private corporations to challenge scientific evidence on a host of contemporary issues. In the early years, much of the money for this effort came from the tobacco industry; in later years, it came from foundations, think tanks, and the fossil fuel industry. They claimed the link between smoking and cancer remained unproven. They insisted that scientists were mistaken about the risks and limitations of SDI [the Strategic Defense Initiative, or Star Wars]. They argued that acid rain was caused by volcanoes, and so was the ozone hole. They charged that the Environmental Protection Agency had rigged the science surrounding secondhand smoke. Most recently - over the course of nearly two decades and against the face of mounting evidence - they dismissed the reality of global warming. First they claimed there was none, then they claimed it was just natural variation, and then they claimed that even if it was happening and it was our fault, it didn't matter because we could just adapt to it. In case after case, they steadfastly denied the existence of scientific agreement, even though they, themselves, were pretty much the only ones who disagreed.

The way that these scientists managed to get their views, and the appearance of a lack of consensus among scientists, was by exploiting journalistic balance required until the 1949 Fairness Doctrine. Under the doctrine, journalists in the U.S. are required to dedicate airtime or column space to present issues in a balanced manner, such as by presenting both sides of an argument. This sounds good in theory, but when one side of the argument consists of hundreds (or thousands) of scientists and the other side consists of a handful of naysayers, it is hard to see how this doctrine doesn't just reverse the bias in favour of the minority. And it was effective. The authors note (p.32):

...their ability to invoke the Fairness Doctrine to obtain time and space for their views in the mainstream media was crucial to the impact of their efforts.

This allowed these scientists to create doubt in the minds of policy-makers and the public, making it difficult for these groups to assess the veracity of scientific claims (and counterclaims):

"Doubt is our product," ran the infamous memo written by one tobacco industry executive in 1969, "since it is the best means of competing with the 'body of fact' that exists in the minds of the general public". The industry defended its primary product - tobacco - by manufacturing something else: doubt about its harm. (p.34)

The book goes through a narrative that is mostly chronological, with chapters devoted to tobacco from the 1950s to the 1970s, strategic defense (the Star Wars programme in the 1980s), acid rain and the ozone hole in the 1980s and 1990s, secondhand smoke in the 1990s, global warming from the 1990s onwards and the recent (at the time of writing the book) re-opened debate about DDT (a pesticide banned in the 1960s). Oreskes and Conway conclude that:

The link that unites the tobacco industry, conservative think tanks, and the scientists in our story is the defense of the free market.

Throughout our story, the people involved demanded the right to be heard, insisting that we - the public - had the right to hear both sides and that the media had an obligation to present it. They insisted that this was only fair and democratic. But were they attempting to preserve democracy? No. The issue was not free speech; it was free markets. It was the appropriate role of government in monitoring the marketplace. It was regulation. (p.248)

That may well be the case, but I was surprised that the authors singled out Thomas Schelling as one of their bad guys (in the climate change chapter), for identifying the importance of uncertainty in climate projections. And I was even more surprised that they also took issue with William Nordhaus:

What could be done to stop climate change? According to Nordhaus, not much. The most effective action would be to impose a large permanent carbon tax, but that would be hard to implement and enforce. (pp.178-179)

Later on, Oreskes and Conway go on to say:

A handful of economists in the late 1960s had realized that free market economics, focused as it was on consumption growth, was inherently destructive to the natural environment and to the ecosystems on which we all depend. (p.183)

One of those economists? William Nordhaus, though the authors don't note this. [*]

That gripe aside, this book was a really timely read, given the post-truth world we are currently living in, where it seems that even the most robust scientific findings are made to look questionable. If this book had been written now, I wonder if this passage would be even need to be re-written:

With the rise of radio, television, and now the Internet, it sometimes seems that anyone can have their opinion heard, quoted, and repeated, whether it is true or false, sensible or ridiculous, fair-minded or malicious. The Internet has created an information hall of mirrors, where any claim, no matter how preposterous, can be multiplied indefinitely. And on the Internet, disinformation never dies. (p.240)

For some more on this, I recommend this Tim Harford post (which, you will probably note, covers some of the same ground as the Merchants of Doubt book).

Overall, despite the minor gripe about Nordhaus and Schelling, I found this an excellent read. It also makes me think about the current situation in New Zealand with alcohol licensing and local alcohol policies, where the industry has funded some limited research. Food for thought.

*****

[*] This bit made me angry enough that the next book I am reading is Nordhaus's The Climate Casino. You can expect a review of that book in due course.

Scale vs. scope economies and the AT&T-Time Warner merger deal

I've been watching with interest the unfolding news on the proposed merger between AT&T and Time Warner. The Washington Post has a useful primer Q&A here:

AT&T, the nation's second-largest wireless carrier, is buying Time Warner, the storied media titan that owns HBO, CNN and TBS. In an unprecedented step, the deal is going to combine a gigantic telecom operator — which also happens to be the largest pay-TV company — and a massive producer of entertainment content.

In ECON110, one of the topics we cover includes media economics. The economics of media companies is of interest because it illustrates a bunch of economic concepts, and because their interaction leads to some seemingly-counterintuitive real-world outcomes.

For instance, media content (e.g. movies, television shows, music albums) is subject to substantial economies of scale - the average production cost per consumer of media content falls dramatically as you provide the content to more consumers. This is because the cost of producing content is relatively high, while the cost of distributing that content (especially in the digital age) is extremely low. Large economies of scale (in distribution) tend to favour large media companies over small media companies, since the large media company can distribute to a larger audience at lower cost per-audience-member. In other words, each item of media content gives rise to a natural monopoly.

However, consumers demand a variety of media content, and variety is difficult (and costly) to produce. Every different item of content requires additional scarce inputs (e.g. quality actors, sets, scripts, etc.), and the more novel the item of content the more expensive it will generally be to create. This leads to what is termed diseconomies of scope - the more different items of content that a media company produces, the higher their average costs. Diseconomies of scope (in production) tend to favour smaller media companies that focus on specific content niches.

So, the combination of these (economies of scale and diseconomies of scope) leads to an industry that is characterised by several large media companies producing 'mainstream' content (and usually formed by mergers of previously smaller media companies), as well as many smaller niche providers. If we focused only on the economies of scale, we would be left wondering why the smaller providers continue to survive alongside their larger rivals. It also explains why the offerings of the large media companies are often pretty bland when compared to the offerings of the smaller players - producing something that isn't bland is too costly.

Which brings me to the AT&T-Time Warner merger. This introduces a new element into play. Time Warner is a content provider, and a distributor through traditional media channels (television, etc.). AT&T is not a media company (at least, not yet), but would greatly enhance the potential distribution of Time Warner content. That in itself isn't problematic, and would simply follow previous media mergers focusing on increasing the gains from economies of scale.

The main problem is that AT&T is a gateway to consumers (and a gateway from consumers to content), and that means that the merged entity could significantly reduce competition in the media market. Media consumers value variety (as I noted above), but this merger might make it more difficult (or costly) for AT&T consumers to access that variety. The WaPo article linked above notes:

Here's where it starts to get really interesting. AT&T could charge other companies for the rights to air, say, "Inception" on their networks, or for the use of the Superman brand. Left unchecked, AT&T could abuse this power and force other Web companies, other cable companies, other content companies or even consumers to accept terms they otherwise would never agree to.

One of the things that AT&T might also do is make it costlier for their subscribers to access other content. Of course, they wouldn't frame it that way - they would instead offer a discount on accessing Time Warner content (which effectively makes other content more expensive and shifts consumers towards the Time Warner content).

What happens next? Federal regulators are looking into the deal, and depending on who you read, President-Elect Trump will either block the deal (which markets are anticipating and which he noted during the campaign) or not (based on the makeup of staff within the administration). From the consumers' perspective, let's hope that Trump follows through (this may be the only Trump campaign promise I would make that statement for!).

Streams aren't drowning music sales

A couple of weeks back I wrote a post about digital music sales, specifically about Radiohead's short-lived free offering of their album In Rainbows, and how it didn't impact on their digital music sales. A broader question, that we discuss in ECON110, is how new business models are affecting more traditional revenue streams in the media markets. Take the example of streaming music (or streaming video), which has exploded in popularity in recent years (although not everyone believes this is necessarily an example of a new industry - witness the notorious views on Spotify of Radiohead's Thom Yorke).

The important question, from the point of view of the artists, is does streaming pay off? Does it cannibalise music sales (digital or otherwise)? A recent paper (ungated version here) by Luis Aguiar (Institute for Prospective Technological Studies in Spain) and Joel Waldfogel (University of Minnesota) looks specifically at this question.

There are third main possibilities with digital streaming. First, they might cannibalise music sales, because listeners find streaming cheaper or more convenient than owning their own catalogue of music. Second, they might stimulate music sales, perhaps because they allow people to sample music they otherwise would not have heard. Third, they might increase revenues despite having little effect on music sales, because consumers who previously downloaded the music from illegitimate sources instead decide to stream the music. Or some combination of these.

The biggest problem with looking at this econometrically is that if you simply run a regression with music sales and streaming, you are likely to observe a positive relationship, simply because more popular music tracks will attract both more music sales and more streaming. You can tell a similar story about pirating of music. Indeed, when Aguiar and Waldfogel look at song-level data, they observe positive relationships.

So, instead they move to looking at aggregated data, which will be less vulnerable to song-level popularity effects. Specifically, they use weekly data on song-level digital sales in each of 21 countries for 2012-2013, artist-level piracy via torrents, both aggregated to the country level, and a country-level index of Spotify use. They find several important results:

First, when we use aggregate data, we find sales and piracy displacement by Spotify. Second, when we use data on the US covering the period of substantial change in Spotify use, we find smaller sales displacement than when we use 2013 data. Third, the coefficients relating Spotify use to recorded music sales differ by format. The coefficients for digital track sales are generally higher than the coefficients for albums, but many of the album coefficients are also negative and significant. Our best estimate indicates than an additional 137 streams displaces one track sale.

Now, that isn't the end of the story. Artists should be less worried about the degree of cannibalisation per se than what it does to their revenue. Streaming adds to artist revenue, while lost sales obviously reduce revenue. What is happening to revenue overall? Aguiar and Waldfogel show that, based on reasonable assumptions about revenue per stream, that Spotify is essentially revenue-neutral for artists.

Finally, there are a few reasons to be wary of these results. First, as Andrew Flowers has noted, the per-stream artist revenue that Spotify claims is contested. So, perhaps it is lower and leads to net decreases in artist revenue. Second, the results are based on data from streams of top-50 songs, which are extrapolated to all streams. There is no way to know whether this extrapolation is reasonable without access to Spotify's proprietary data (and they're not volunteering it - if I was cynical, I might point out that this is awfully convenient!). Third, as Andrew Flowers points out, we don't know about the distribution of artist revenue. It's likely to be concentrated among a few high-profile artists, with the little guys getting very little. So, streaming might be net positive for the big artists and net negative for the little artists (or the other way around), and net neutral overall. Again, this is something we couldn't evaluate without more disaggregated Spotify data (and a good identification strategy to overcome the issues noted above).

[HT: Marginal Revolution]

Where Radiohead went right, and Nine Inch Nails went wrong

In ECON110 every year we have a half a topic on the economics of media industries. Part of that topic is considering new revenue stream models for the media industries that are most 'at risk' from digital downloads (i.e. those industries that are most reliant on revenue from content, as opposed to revenue from selling audiences to advertisers).

One thing I always cover is the Name Your Own Price (NYOP) movement in the recording industry. There are a number of notable examples, like the Canadian singer-songwriter Jane Siberry, who allows downloaders to choose the price they pay for any song they download - if they choose to pay nothing, it is "a gift from Jane" (although I note now that her site is down?). Other examples include Radiohead's almost-free pre-release of their 2007 album In Rainbows, where Radiohead released the album on their website where fans could pick their own price. The pre-release was available from October 10 to December 10, 2007, after which the album was released in digital and physical forms in early January 2008. The pre-release was 'almost-free', because even if someone was to choose to pay zero, they had to pay a service charge of £0.45, plus they had to provide their contact details (a non-monetary cost).

A paper earlier this year (ungated earlier version here) by Marc Bourreau (CREST-LEI), Pinar Dogan (Harvard) and Sounman Hong (Yonsei University) looks specifically at the impact of Radiohead's strategy. Specifically, they look at the possibility of two effects occurring: (1) a 'cannibalisation effect', whereby the early online sales reduced later sales (that would have occurred had the pre-release not happened); and (2) a 'buzz effect', whereby the pre-release generated a lot of media attention, increasing the profile of the album and increasing overall sales.

They compare the release of In Rainbows to the releases of a control group of bands similar to Radiohead, based on a variety of criteria (the control group was Sigur Ros, Muse, Arcade Fire, Beck, Interpol, Bjork, Coldplay, The National, Arctic Monkeys, Franz Ferdinand, and Sonic Youth). They find:

In the case of conventional album releases, the estimated coefficient at the "peak" is about 2 for digital album sales... This is translated as an increase in the digital album sales, which are about 7.4 times higher in the week of the release compared to the previous week...

In the case of the PYOP pre-release strategy of In Rainbows, the coefficient in the "peak" is estimated to be about 3.7 for Radiohead's digital albums. This estimate suggests that Radiohead enjoyed an increase of about 40 times in its digital album sales compared to the week before.

It's easy to see that there is a substantial difference between a 3.7 times increase in sales, and a 40 times increase in sales. On the flip side, there was little impact of the digital pre-release at all on CD sales (they neither increased nor decreased as a result). The authors conclude that the pre-release strategy generated a substantial 'buzz effect' for Radiohead, that more than offset any 'cannibalisation effect' on digital sales. They provide additional evidence for this in the form of Google Trends searches for Radiohead, and mentions of Radiohead in print articles (from Factiva).

What about Nine Inch Nails? They released their 2007 album The Slip for free download from their website (it's still available there now). There wasn't nearly as much hype about this release as for Radiohead (according to Google Trends and Factiva), and Bourreau et al. find that the release had a negative effect on NIN's digital album sales (when compared with their control group: Marilyn Manson, A Perfect Circle, Tool, Queens of the Stone Age, Deftones, KoRn, Smashing Pumpkins, Massive Attack, Rob Zombie, and The Prodigy). In other words, the cannibalisation effect outweighed the buzz effect.

Overall then, a win for Radiohead (in more ways than one - In Rainbows won two Grammy awards). NYOP works, but only if you can generate enough hype to ensure that sales increase overall. Or, it might work also for new artists who are trying to overcome the information asymmetry problem (potential listeners have no idea if they are any good or not) - this is the model that bandcamp.com is founded on.