What happened to newspaper revenue? This short post contains an answer in two graphs. The graphs show the growth and decline in aggregate newspaper revenue.
More to the point, the graphs illustrate the timing of the decline in 
revenue and also the acceleration in decline. The graphs provide useful information about changes in the composition of revenue, which helps explain why revenue declined.
The graphs come to this space from my esteemed colleague, Tom Hubbard, who is the John L and Helen Kellogg Professor of Management & Strategy. Tom teaches a class on advanced strategy for MBAs, and had compiled the data for a module in the class.
In brief, Tom read a previous post in this space about the decline in newspaper revenue, and we got to talking about how a small amount of well chosen data can illustrate insightful analysis. He showed me what he had collected and I was impressed. He graciously offered the graphs as additional evidence for the conversation.
What do the data say?
Consider this following graph. It shows total ad revenue for US newspapers over the last two decades. (Advertising revenues are Newspaper Association of America estimates, based on quarterly earnings of newspaper publishers, accounting for half of U.S. circulation, projected to the universe of U.S. daily newspapers.)
Notice: This is ad revenue, not subscription revenue. Why examine that? Because it focuses attention on the major source of revenue for newspapers.
Two features of the picture should get your attention. First, the peak year for newspaper revenue was about 1999-2000, with a noticeable fall after that. The second feature is related to the first one. Newspaper revenue remains roughly the same for the five years 2000-05. Then revenue starts to decline again, accelerating into a (virtual) nosedive thereafter.
The patterns in the graph frame two questions. What happened in 2001 and what happened in 2006?
The next graph helps address both those questions. It decomposes the total revenue for newspapers into four components: Classified ads; retail ads (to local advertisers); national ads (for national newspapers); and online ads.
Take a look at this second graph. What do you see?
The first observation should be obvious. The decline in 2001 was almost entirely due to the drop in classified ads, with a mild contribution from retail ads. Two explanations are likely: Perhaps that was partially due to the recession, which accelerated after the dot-com bust and 9/11, and partially to the growth of the Internet, which, by that point, had spread to approximately half of all US households.
The second observation is more surprising. Classified ads have been in steady decline since 2001, with a noticeable acceleration in the degree of decline after 2005. Only a brief growth in online ad revenues after 2001 made up 
for those declines in classified ads between 2001 and 2004. Yet, growth in online ads never accelerated after 2004, and, accordingly, it did not serve as a counter force, as so many newspapers had hoped for only a few years earlier.
The second graph nicely relates to something mentioned in the last post. Specifically, the acceleration in the decline of classified ad revenues after 2004 is largely due to Craigslist. The graph shows this better than words can describe. There is an enormous shrinkage after Craigslist goes into overdrive in the middle part of the decade, spreading to hundreds of cities throughout the United States at this time.
Finally, the second graph illustrates one last point. Around 2005 retail ads also begin to decline at an accelerating pace. That decline has a very different origin. It reflects declining newspaper readership, as more readers move their time online to other sources of news. That trend is quite distinct from the one shaping classified ads.
Summarize: The acceleration after 2005 is, therefore, the result of several factors. It came from (1) the decline in ad classified revenue; (2) the decline in readership; and (3) the lack of growth in online ad revenues.
That also frames one of the most interesting underlying questions in this topic. How did (2) and (3) happen together? As readers went online didn’t they become targets for more efficiently delivered ads? Nobody would expect a one-to-one match of declining newspaper ads and growing online ads, but it would be sensible to expect decline in one to lead to some growth in the other. Why didn’t the decline in ads in newspapers get matched by some growth in online ad revenue?
Ah, there is a good question for another day.
Wrap up
Isn’t it interesting how one graph simultaneously can illustrate many points? That is a very informative graph, don’t you think?
More to the point, I bet Tom Hubbard’s class on advance strategy is a very insightful class. Thanks for sharing the data, Professor Hubbard!
************************************
Late post script: Joshua Gans writes in with an excellent point about a missing explanation for the decline in revenue at newspapers, especially recently. Not only are newspapers getting fewer readers in the last few years, but the remaining readers are less valuable readers, as competition for their attention increases. Quite a relevant point, and a good start on the open question posed at the end of the post. If you have a taste for economic theory about the competition for attention, feel free to read more from Gans and coauthors Athey and Calvano.
Virulent Word of Mouse 
Having worked at the LA Times, I’d like to add one additional layer of complexity to this analysis – though this information is not apparent in the graphs from Professor Hubbard’s class. Advertising revenue is defined by Ad Space (or number of ads) x Ad Rate x Circulation. Even though newspaper circulation has been in decline for well over a decade, newspapers have been able to mask this decline because they were able to continually raise ad rates. In a progressively fragmented media market, newspapers were still the best game in town for mass reach in local markets. However they reached a tipping point in 2004 due to a number of factors:
First, due to fragmentation of media, many advertising clients were developing database/direct marketing initiatives. Accelerating ad rates in the newspaper industry were one more nail in the coffin for traditional media. Overnight many annual newspaper ad budgets were slashed by 75% or more in order to pay for expensive direct marketing campaigns.
Second, there was substantial corporate consolidation in some industries (cell phones, fashion retail, et al) that had traditionally been heavy newspaper advertisers. Less industry competition reduced the need for heavy advertising; plus companies were cutting ad budgets to provide the savings efficiencies that justified the acquisition.
Third, and probably greatest of all, was the realization that newspapers were no longer reaching a younger demographic. Newspapers have always skewed older, but they were no longer bringing in younger readers to replace older ones. Their readership was literally dying off.
These three issues prevented newspapers from continuing to increase ad rates faster than the rate at which their circulation was falling, and caused advertising revenue to start to decline at a much more rapid pace than had been seen previously.
Comment by Greg Peterson — August 4, 2011 @ 9:58 am |
Those are very good comments, thanks. Your third comment raises a really interesting question. The following was released on Aug 1, and seems not to agree with your third point.
http://adage.com/article/adagestat/affluent-americans-print-media-tops/229002/
They seem to think the young and affluent experiment, but still go to the newspaper. Great question.
Shane
Comment by Shane Greenstein — August 4, 2011 @ 11:33 am |
It is an interesting study. I would enjoy seeing their survey methodology. I don’t disagree with the findings, but the findings may not mean what they think they mean 🙂
Reservations I have about the study:
(1) It only surveys affluent households. Granted, it was a study designed to understand media use among affluents, but I would be interested to see the related demographics of age, sex, ethnicity and education.
(2) It appears they focused on media use (in general). I would be curious to see media preference, frequency of use, and depth of use.
(3) I’m a little confused about the conclusions they are drawing from product “form”. For example they found that 39% of people surveyed “read” newspapers on their computer. Does this mean they subscribe to an electronic version of the paper? Or just occasionally view a newspaper article online? Do they use the newspaper web site as their primary portal for news? Or are they directed there from another news aggregator (Yahoo! etc)?
In study after study we found that education level is the single greatest predictor for whether a someone reads or subscribes to a newspaper. Income and education are often linked, but in cases where education was high and income low, newspaper usage was still strong, while in high income, low education cases, newspaper use was low (all other things equal).
Additionally, people continue to say they “read” the newspaper long after it ceases to be their preferred source of news. They first stop reading as deeply, then as frequently, and finally they stop reading it altogether and subscribe to it out of habit. We would often ask people if they read the newspaper and they would say “yes”, and then we would ask the last time they read it and it would be weeks prior.
Finally you cannot ignore age in ANY media study today – particularly when comparing traditional media with new media (and new media forms). To ignore age is to ignore the future – as well as to ignore the target audiences of the majority of consumer product companies. I will guarantee you that even within affluent households, media use skews strongly based on age.
Comment by Greg Peterson — August 4, 2011 @ 1:05 pm
*EDIT* My last comment seem to suggest I didn’t see the last section of the study. I should probably rephrase and say – to look at media use in a snapshot does not capture how media use is evolving in younger households. Many younger households learn traditional media use from their parents and “evolve” away from it over time. If you looked deeper into media use segmented by age, you would find very different dynamics at play than those listed in this study.
Comment by Greg Peterson — August 4, 2011 @ 1:17 pm |
Great comments. Thanks.
Comment by Shane Greenstein — August 4, 2011 @ 2:17 pm |