There's been quite a bit of junk research of late. First my friends at the Direct Marketing Association proudly presented 'research' claiming that nearly a quarter of all sales in the UK are the result of unsolicited marketing (I'm not making it up); then there was a report from the Office of Fair Trading which dealt with the question whether or not the Yellow Pages still has a monopoly position in the market for unsolicited phone books (its answer is negative); and now we got a study into how much spam is costing the US economy.

The study into unsolicited junk e-mails is The Economics of Spam by Justin Rao and David Reiley. You can find it on the website of the Journal of Economic Perspectives; a publication aiming to fill a gap between the general interest press and most other academic economics journals. That's a noble aim but it does result in typical junk research – albeit interesting junk science.

The finding that has been picked up by various media, including the New Statesman and Huffington Post, is that the 90bn spam e-mails sent across the globe every day costs the US economy $20bn (£12.7bn) per year. Also newsworthy: the spam industry makes a relatively modest annual profit of $200m (£127m). In other words, Rao and Reiley claim that the US economy loses $100 for every dollar spammers earn.

The science of guessing

Estimating how much damage spam does to the economy is rather difficult and involves a fair amount of guesswork. It probably suffices to say that a previous study (by Ferris Research) estimated that the worldwide cost of spam was $130bn in 2009. Depending on what indicators you use and what assumptions you make you're easily a hundred billion dollars out.

Rao and Reiley arrived at the figure of $20bn using two indicators. The first is the cost of anti-spam technology and hardware, including the labour cost associated with installing and maintaining anti-spam solutions and the extra server capacity needed to cope with spam e-mails. This cost is estimated to be $6.5bn. The article doesn't tell us much about how this has been calculated; all we're told is that it's based on the above-mentioned work done by Ferris Research and that the authors deem the figure to be a reasonable estimate.

The second indicator is the number of hours lost by people dealing with spam. This is where the research becomes unconvincing. I actually think the researchers realise this as they fail to show how they've arrived at a cost of just under $14bn. Once you do the calculation yourself you start seeing it's flawed.

Time to get your number cruncher out…

• First, the researchers guestimate (they prefer the term educated guess) that 1.2% of the 90bn daily spam messages end up in our inboxes (most junk e-mails are weeded out before they reach our e-mail accounts). This gives us 1,080,000,000 spam e-mails per day, or 394,200,000,000 per year.
• They then estimate that deleting a single spam e-mail takes about five seconds. So, deleting 394,200,000,000 spam e-mails would take 547,500,000 hours ((394,200,000,000 x 5) / 3600).
• Finally, they assume that the average wage in the States is $25 per hour. You can then work out that deleting a year's worth of spam costs (547,500,000 x 25 =) $13,687,500,000.

As I said, this is junk research. Not only is it based on guestimates about how many spam e-mails end up in our inboxes, it also takes for granted that all 90bn spam e-mails sent every day are targeted at people living in the States. Even worse, the researchers wrongly assume that you can attach a cost to individuals deleting a handful a spam e-mails a day. Such assumptions are somewhat idiotic. Just consider this:

• According to the International Telecommunications Union there are exactly 243,542,822 internet users in the US. Assuming that all these people use e-mail we can calculate that the average internet user in the States 'only' gets 4.4 spam e-mails a day (1,080,000,000 / 243,542,822).
• Deleting those messages takes the average person 22 seconds per day – at a cost to the US economy of 15 cents.

That indeed is as ludicrous as it sounds. The assumption that the 22 seconds would otherwise have been spent productively and earned another 15 cents for the US economy is nonsense. It assumes that people are productive all the time apart from those few seconds spent on deleting junk e-mails. Attaching a cost to such things as deleting a handful of spam e-mails is plain stupid. It's as daft as arguing that the US economy would grow by many billions of dollars per year if only everybody would be willing to work a couple of seconds per day longer. Or how about calculating how much blowing your nose costs the economy?

How to tackle spam

Apart from the poor attempt at quantifying the cost associated with spam the article Rao and Reiley have written is excellent. It contains a well-written history of spam and contains lots of interesting observations. Naturally, I quite liked the comparison between spam and junk mail. What they got in common is that they impose 'negative externalities' upon the consumer. That is, other types of advertising may be annoying but at least we get something in return, such as free online services and affordable newspapers and magazines. Spam and junk mail (as well as unsolicited phone calls and bill boards) don't benefit us in such a way. On the contrary – instead of getting free or affordable services we're presented with a bill for disposing of the pests.

And yes, spam is a greater evil than junk mail. Firing off millions of spam e-mails is so cheap that a response rate of 0.0001% can be enough for the sender to break-even. Compare that with ordinary junk mail, which needs a response rate of roughly 2%. In other words, the reason why spam is such a major problem is that there's so much of it. If it were not for the organisations fighting spam we would have stopped communicating via e-mail a long time ago – marketeers would flood our inboxes with unsolicited e-mails to such an extent that nobody would want to go anywhere near an inbox.

Equally interesting is the article's discussion about what governments could do to tackle spam. One of the concepts discussed are 'attention bonds'. The idea is that the sender of an e-mail pays the recipient a small sum of money for 'attention'. If the recipient opens an e-mail the money is automatically transferred from the sender to the recipient - although the recipient has the option to accept the e-mail without payment. This would obviously make marketeers very reluctant to send spam but, sadly, the idea doesn't seem feasible. Linking all e-mail accounts to bank accounts is not without risks…

More realistic solutions, according to Rao and Reiley, is attacking the supply chain. This could be done by taking against banks that facilitate spammers and by making spam less profitable by placing lots of fake orders on the websites spam e-mails redirect people to.