Ignorance and Competition in the Book Market

I’ve been having a long Twitter discussion with Will Wilkinson about the economics of the book industry. Will wanted to know how authors could make money without “digital rights management” technology, and I replied by saying that writing a book has always been less about making money than it is about promoting the book’s author, and I suggested that the activity is going to become a lot less lucrative in the future.

To think clearly about the future of the book industry, it’s essential to understand its present. In particular, to make an educated guess about what the price of a book will be in the future, we need to understand why books cost what they cost today. I think there are two key insights that largely explain the structure of the book market. First, publishers are really bad at predicting whether any given book will be a success. And second, there is an almost unlimited supply of aspiring authors, some non-trivial fraction of whom would, if given the opportunity, produce a best-selling book. Call these the ignorance and competition assumptions, respectively.

Now, today’s book prices have two distinct and puzzling characteristics. First, book prices don’t vary much and don’t seem to be correlated with popularity or quality. Most categories of hardcover books cost around $25.00. Publishers do not seem to charge extra for books by famous people. They don’t jack up the price if a book is well-reviewed, or offer discounts they’re panned in the New York Review of Books. As Paul Graham has observed, publishers seem to price their books in rough proportion to the cost of the raw materials: longer books are somewhat more expensive, but what’s actually printed on the page has relatively little effect on a book’s price.

This can be explained by ignorance (on the part of customers) and competition (among publishers). A customer generally don’t get to read a book before you buy it, and other indicators of quality, such as reviews, are only weakly correlated with a given customer’s enjoyment of a book. Moreover, there are lots and lots of books to choose from. These factors, together, make the book-buying public strongly price-sensitive. A publisher selling a $40 book in a market where the norm was $25 would lose a lot of customers. Because book-buying is always a hit-or-miss affair, few would want that specific book enough to pay an extra $15 for it, while most would have plenty of other books of similar perceived quality and dramatically lower price.

The second puzzling characteristic is that books are dramatically more expensive than their cost of raw materials. Printing and distributing a book costs around $5. This means that every book sold for $25 represents a huge profit to be divided among the bookseller, publisher, and author. This too is explained by ignorance, this time on the part of publishers. The print process is characterized by high fixed costs and economies of scale. This means you have to sell several thousand copies of a book to recoup the costs of printing it. Most books do not hit this target and so lose their publishers money. Hence, when you buy a book for $25, you’re not only covering the $5 it cost to print and distribute that book, but you’re also helping to defray the costs of several other books that wound up in the remainder bin.

With this background, it should be easy to see how authors’ compensation is determined. To simplify the math a bit, let’s assume there are only two outcomes for a book: hit or not-hit. Then the value of a manuscript is the profit from a hit, times the probability of a hit, minus the losses from a non-hit times the probability of a non-hit. Our ignorance assumption means that the probability of a hit is low, which means that the expected value of printing a book—and hence the value of a manuscript from a first time author—is very low. And this is what we see in the marketplace. When a publisher takes a chance on a non-famous, first-time author, the advance tends to be relatively small.

Things look different for repeat authors because the ignorance assumption doesn’t apply with the same force. Stephen King and J.K. Rowling have demonstrated that they can write books that appeal to large audiences. And this means that not only is the payoff for a hit higher, but the probability of a hit is much higher as well. And this puts them in an extremely strong bargaining position with publishers and allows them to become very wealthy. Hence, we see a highly skewed distribution of earnings, with a tiny minority of authors getting multiple hits and making millions of dollars, while a huge number of authors write only one or two books, get paid very little for it, and fade back into obscurity.

Now imagine a world with omniscient publishers. Every publisher can now predict exactly how many books any given author will sell. This will have two effects. First, obviously, publishers will no longer print money-losing books. Only those books that can recoup their costs will be printed. And second, given my competition assumption, many more best-selling authors will be discovered. Both of these developments will push prices downward. The ability to avoid wasting money on duds means that publishers have a lot of room to cut prices. And publishers’ ability to find new bestselling authors greatly increases the number of bestselling books that can be printed. A world of omniscient publishers would be a world of commodity publishing: publishers would get much smaller margins and bestselling authors would get much smaller advances.

So let’s return to our own, non-omniscient world. We might say that what makes a best-selling author valuable isn’t just his writing talent—an author was probably just as talented before he was discovered as after—but in the knowledge that the author is, in fact, capable of producing best-selling books. And producing this knowledge is (or at was until recently) really expensive—to find one John Grisham, you had to publish a bunch of books by unknowns and see which ones sell. This means that the people who have convinced a publisher to bear the costs of “discovering” them have what amounts to a uniquely valuable credential. They can extract significant rents because even though there are likely plenty of others who could produce novels of similar quality, it’s too expensive to figure out who they are.

In my next post, I’ll explore how these considerations shake out in the age of the Kindle.