A tax on targeted ads?

When Paul Romer expresses an opinion, it is always worthwhile to listen because it is always well-considered. In an opinion piece in the New York Times, he puts forward a proposal to restore what he terms is the “public commons” of the provision of information in support of democracy. He actually puts forward two linked proposals: one for a target on targeted ads by digital platform companies and a proposal that the tax is progressive (which may be a check on dominance). The latter is interesting but I will just focus on the former here.

Romer opines that it is the ad-based business model that is at the heart of problems of misinformation on Google and Facebook and that we would be better off with a subscription-based model. By taxing ads (or ad revenue), you would cause these platforms to shift towards subscriptions. There are lots of issues here so let me try to disentangle them.

First, what is wrong with the ad-based business model? Romer argues that it encourages content that works as ‘click-bait.’ In other words, it attracts attention in a way that encourages people to see more ads but not in a way that might be more informative — say with deeper exposition and understanding. The point is that attention is limited and we should worry about how people allocate their attention.

But are ads the problem? It is hard to know. Certainly, it is a reasonable supposition that, if you pay for some content, you will intend to pay it more attention. If you buy a book, you probably intend to read it someday. If you subscribe to The Economist, ditto (even if there are ads there). The problem is that the more time you spend on the actual content, the less attention you have to devote to the ads.

That is, of course, just one effect. We can’t just look at subscription models and, say, “see, everyone should consume content that way.” We have to consider the counter-factual. What would the people who do not currently pay for content do, if ad-based models weren’t available? Some may move towards a more ‘informed’ mode. But others — and let’s face it — probably most, would not. Their attention would go elsewhere. Put simply, a starving kid may eat their veggies but they may also go outside the system when you aren’t looking.

Second, Romer focusses on platforms that engage in ad-targeting. This suggests that he doesn’t think the business model is a problem for media that just throws up ads without regard to whether they know the individual characteristics of consumers or not. Ad-targeting tends to allow more revenues to be earned by platforms so by taxing it, you are taking the cream of the crop. But why does it have those more revenues? For advertisers, they are willing to pay more for ads if they know they will actually grab consumer attention. But I suspect — given my analysis above — that it is this aspect that Romer sees as the problem. A successful ad takes away from the consumer looking at deeper content. So it may be a vicious circle.

That said, the reason consumers are paying more attention to those ads is that they are more engaging, useful and about things they want to buy. In other words, they aren’t just wasted stuff you have to sit through on broadcast TV but instead are things you might find more useful. I know that everyone claims they never click on ads. The evidence is that is not true and it certainly isn’t true for me. If I have a choice, I’ll have ads that something thinks might be useful for me. Moreover, my cursory glance at ads these days, compared with the Internet of old, looks far less ad-cluttered than before. This is likely because serving too many ads caused consumers to go elsewhere. The social media of today really has a gentle number of ads. What I don’t know is whether ad-targeting has helped make that happen or not.

A tax would divert away from this stuff. What exactly? Ads that take into account individual information before they are served up. In other words, they are not generic. Romer is at pains to consider whether businesses might get around this. He means in terms of paying the tax and argues that if they stop showing ads and don’t pay tax, then great. But is that the only option. Suppose I want to target people in a certain location, who like baseball then I could use ad-targeting, or I could place an ad in the sports section of the local paper/website. Maybe this would encourage more local news but my point is that you can tailor content to attract a certain type of person. It is harder than ad-targeting but it has been done since advertising was invented. And if that is the case, will the tax really be effective?

In the end, we seem to be a long way from being able to properly evaluate a proposal like this. I personally do not have a clear picture as to the effect this would have on digital platforms and the supply of better information. And even if that picture is gained, there is the empirical matter of whether all the likely ducks will line up in a road for this proposal to work. It is an interesting idea but we should read it as a proposal for more research than a proposal for action.