Nothing succeeds like success---or why "you earned it all on your own" doesn' quite cut it in the real world

Crooked Timber:

The discussion in the comments section to my last post has turned to the old disagreement over the respective contributions of innate quality (however you want to try to measure it) and external circumstances to economic ‘success.’ Matthew Salganik and Duncan Watt’s research on cultural markets provides some very interesting insights into this question. This paper is the best overall survey of their findings that I know of. What they do is to set up a set of discrete artificial cultural markets, in which large numbers of experimental subjects listen to pop music, and rate it for whether they like it or not. All of these markets have the same music. Most of them have an effective recommendation system (in which subjects can see which are the more popular, and which the less popular tunes); two do not provide this information. Subjects are randomly assigned to different markets.

Salganik and Watts do find that a song’s ‘appeal’ (they are reluctant for obvious reasons to talk about innate quality) has consequences – ‘more appealing’ songs tend to do better. However, this is a pretty noisy process – and is especially noisy in markets where people can see which songs other people liked the most. These markets are both more unpredictable and more unequal.

Overall, the results from Experiments 1 and 2 provided strong support for the argument that social influence at the individual level is simultaneously responsible for increased inequality and unpredictability in collective outcomes—in this case, the distribution of market share. Although simple to state, this finding nevertheless exhibits a curious paradox: On the one hand, by revealing the existing popularity of songs to individuals, the market provides them with real, and often useful, information; but on the other hand, if they actually use this information, the market inevitably aggregates less useful information. This result, which is analogous to ‘‘information cascades’’ in economics (Banerjee, 1992; Bikhchandani et al., 1992), suggests, in turn, that social institutions that make us aware of the behavior of others—the New York Times bestseller list, the Billboard album charts, and lists of top-grossing movies—do provide a useful service to individuals, but only at the cost of increasing the overall inequality and unpredictability of the markets themselves.

This may seem at first not to have implications outside pure cultural markets, but I suspect that it does have more general implications for any market in which people are able to observe each other’s market choices (which is to say, nearly any market you can think of – think about Keynes’ discussion of animal spirits and investors here). In particular, I think that it has substantial implications for employment markets. Here too, potential employers ‘observe’ both each other’s choices and the choices made by non-employers while choosing employees. They are more likely to hire employees who appear to have done well in other positions, to have gone to ‘good’ universities etc. The point is twofold. First – that this can create substantial path dependence (Salganik and Watts discuss the congruence of their findings with Arthur’s). People who do very well at the beginning of their careers (perhaps through sheer luck; perhaps not) are more likely to keep on doing well, precisely because new opportunities open up to them which would not otherwise have been available, perhaps themselves leading to new opportunities etc etc). The second (which is forcibly born home by a very large body of sociology) is that these early opportunities (as well as other, later ones) are not randomly generated, in contrast to the advantages that fortuitously arise in Salganik and Watts’ experiments. Many of the initial informational signals that are sent out (attendance at good universities; prestigious internships) are disproportionately available to those who come from well-off backgrounds. And those who start out with advantages tend to end up doing better. Again, ability, or quality, or ‘appeal’ or whatever has real consequences. But it intersects with other, external advantages, which are usually not randomly assigned (and that’s not even to get started on more direct forms of market discrimination).

None of this is exactly surprising – but it is often ignored by those who have done well out of the system as it is. It also applies with especial force to academia. Kieran has an old post that I can’t locate easily (lazyweb, I invoke thee) on exchange practices among graduate programs, and how people from top ranked programs tend to do very well in initial job assignments. Obviously, your first academic job (if you can get one) has important consequences for the rest of your career – if you get a tenure track position at a well ranked program, you are likely to have research and funding opportunities that are unavailable to your less lucky peers, even if those peers are in some sense ‘better’ than you. And these advantages cumulate over time. This is something that my personal history makes me very aware of. I didn’t come from a top ranked program (although I did have some other advantages) – but my initial piece of luck was a two year post doc with minimal formal responsibilities and lots of research resources, which allowed me to write a risky article aimed at a top ranked journal in my field. Getting that early publication gave me exposure, which helped me get a good job, get various research opportunities, take further risks etc – all of which cumulated in the right way. But if I hadn’t had that initial advantage, I would have had a much tougher time of it, and plausibly would have dropped out of academia. And others – who plausibly could have made as good or better use of that advantage if they had it – didn’t get it. The general point is not that ability doesn’t count at all – but that opportunities to exercise ability count too, and that a world where they are cumulative (so that more opportunities come to those who have had such opportunities in the past because of information sharing or another mechanism) is likely to generate high levels of inequality of outcome and substantial ‘noise’ in the relationship between ability and outcome.