Anomalous Presumptions

In a post based on his new book Tyler Cowen recently questioned whether revealed preferences, measured by monetary exchange, are an adequate framework for economic analysis. I’ve been planning to post further on non-monetary transactions and his quote has pushed me over the threshold.

In “The cost of money” I explored why so many network-mediated interactions are non-monetary, but I only looked at the negative side – the irreducible transaction costs of money. Now I want to explore the positive side – what do people get for contributing to open content production, such as testing or writing open source software, writing for Wikipedia, proofreading for the Gutenberg project, blogging, posting pictures on Flickr, etc.? With very rare exceptions contributors don’t get money, but rather are willing to spend money – they often have to pay at least a little for internet service and hosting, and open content production sites seem to get quite a few cash donations when they ask.

Obviously people get many things in return for their investment in content production, including such indirect monetary rewards as better jobs. However I argue that a dominant theme across essentially all kinds of open content production is esteem – the favorable opinion of one’s chosen reference group. We can see the importance of esteem in efforts by contributors to make sure they are properly credited, debates over priority, flame wars over design judgments, etc. Of course making contributions as investments in esteem is not new; we see very similar patterns in scientific and artistic communities going back at least hundreds of years.

Esteem is clearly sought in most cases for its intrinsic value to contributors, not for its instrumental value. For most of us the favorable opinion of our reference group is not a means to other ends, such as higher salaries, it is an end in itself. This is consonant with the recent research (e.g. 1, 2) indicating that happiness, in developed societies, is a function of social relationships and relative economic status, not material well being per se. I think that once most material needs are satisfied, wealth produces happiness largely by helping us to garner esteem.

If we take esteem seriously as an important source of intrinsic value, then we have gone a long way toward explaining how open content production can replace financially mediated content production. For many people, contribution to open content is a very efficient investment with returns in esteem– far more efficient than incurring the double transaction cost of converting work into money, and then money into investments in goods and services that may return esteem.

We are all familiar with the way digital networks are changing the cost of collaboration by reducing manufacturing and distribution costs for information goods asymptotically toward zero. Equally important, given this analysis, is the fact that networks dramatically increase the chance that someone can find a reference group that will esteem the contributions they can make. This is the dual of Benkler’s key point in “Coase’s Penguin” that networks allow for much better recruitment of contributors to projects based on the fit between them. Benkler emphasizes the ability of individuals to make contributions that are valuable to the project; I emphasize the esteem the reference group provides to the individual; but these are two sides of the same phenomenon.

However the replacement of money by esteem is not just a matter of efficiency. We can see this by observing that in many cases, we cannot use money to acquire major sources or indicators of esteem, such as Olympic gold medals, Nobel prizes, or even the laughter of friends. If any of these were known to be for sale they would lose all their value! I have some ideas about why this is true that I will address in a subsequent post.

Filed by Jed at 11:38 pm under Economics
1 Comment

Clay Shirky in 2003 wrote a deservedly famous piece on why micro-payment systems always fail. However Shirky’s analysis stops with the psychological burden of deciding to pay for content. For the reasons noted below, I think this problem has much deeper roots in the nature of (modern) money itself.

First, money is totally fungible and portable – especially virtual money, discussed in more detail below. This makes it an extremely attractive target for fraud and theft, and produces an “arms race” between criminals and those who protect money. Furthermore, because theft or fraud of virtual money can be largely or completely automated, crimes that get even relatively small amounts from very large numbers of people can be attractive (Shirky actually references a description of one technique for this in passing).

Second, the risk of fraud (or simply error) in exchanging money for goods (or services) creates an enforcement cost for each individual contemplating a transaction. Errors are random and would tend to balance out, but the incentive for fraud, and for biasing errors, is great. (These are not really distinct, since fraud is just induced error.) Automated methods make “micro-fraud” feasible and cost effective in some cases. This is the kind of problem that Shirky addresses most directly.

Finally, most modern money is virtual, and exists only as numbers in various databases, and the only thing that keeps money from being created or destroyed is the discipline of the various institutions that maintain the databases. If this discipline breaks down anywhere, money can be created or destroyed by transactions that do not balance. If such creation or destruction becomes common enough, the failure undercuts the fundamental properties that make money work.

This sort of global transaction discipline requires a substantial enforcement effort – and therefore cost – to maintain, especially given the massive incentives to cheat.

Now let’s put these problems in the context of “frictionless” network activities. Exponential declines in the cost of network bandwidth, storage, and computation are driving the cost of producing and distributing many services and information goods toward zero. The costs of processing transactions involving virtual money can ride these same exponential cost curves down toward zero. However the cost of enforcing honest monetary transactions cannot follow the same cost curve. Enforcement cost is constrained by the potential for fraud and error, and we can’t just depend on automated enforcement mechanisms because they are always open to automated theft and fraud.

The result is that as information goods and services become cheaper, they encounter a threshold where their price is so low that the enforcement cost of monetary transactions is greater than the value of the transaction. At this point there are two alternatives: they can stop following the declining cost curve, or they can be transferred through non-monetary transactions. While some goods and services have stopped following the curve, many have become “free” (i.e. supported by advertising and/or voluntary contribution). Advertising is a common example of a non-monetary transaction – the consumer “pays” some attention to the ad in exchange for the good or service. Voluntary contributions are more subtle, but we can consider them transactions in which the consumer “pays” attention to the producer themselves – the way we pay performers with applause and credit.

In one sense this is a normal, economically rational process, and it certainly fits Coase’s general framework of transaction costs. The peculiar thing about this case, however, is that it excludes money itself from the transaction process on fundamental economic grounds, indicating that there are intrinsic problems with using monetary value as a metric for transactions.

Note that as the likelihood of significant loss through fraud or error increases, the minimum value for feasible transactions also increases. It is especially hard to judge the value of contracting for unique goods. Thus in cases of one-time production of information goods such as software, encyclopedia articles, or commentary on current events, the threshold will be much higher than for transactions with frequently replicated outcomes.

Understanding transaction costs in general, and the cost of money in particular, helps a great deal in explaining why so much of the landscape of information goods and services has transitioned so rapidly toward a “free” (i.e. non-monetary) approach. However costs to consumers can only be part of the story. There must also be benefits to producers, and that is a separate (and possibly more important) issue, which I’ll discuss in a later post.

Filed by Jed at 3:52 pm under Economics
1 Comment